The running list is on the regulation tracker.
28 Oct 2026
US Justice Department is asking about Nvidia's Hugging Face deal, as reported by Reuters
Reuters reported from San Francisco on 28 October 2026 that the US Department of Justice has made inquiries about how Nvidia's $12.93 billion deal to buy Hugging Face would affect competition in artificial intelligence. Reuters attributed the account to two people familiar with the matter. It said the questions come at the same time as a years-long investigation into Nvidia's conduct. Reuters said the department's questions about the deal, and details of that broader inquiry, were being reported for the first time.
Source: Reuters, as carried by PiQ Markets (opens in a new tab)
28 Oct 2026
Poland's AI inspection and penalty provisions enter into force
Poland's Act of 3 July 2026 on artificial intelligence systems (Dz.U. 2026 poz. 1003) was published in the Dziennik Ustaw on 27 July 2026. The Sejm's official legal-information page says the act entered into force on 11 August 2026, with two exceptions. Article 125(4) entered into force on 28 July 2026. Articles 8 to 18 and chapters 3, 4, 5, 8 and 9 enter into force on 28 October 2026. Article 127 says those provisions enter into force three months after publication. The 28 October date is the date stated on that official page.
Source: Sejm ELI, Dziennik Ustaw 2026 poz. 1003 (opens in a new tab)
27 Oct 2026
Queensland publishes a bill to criminalise creating intimate images and audio without consent
The Criminal Code (Non-consensual Creation of Intimate Images and Audio Material) Amendment Bill 2026 was published on the Queensland legislation register on 27 October 2026. It is a bill. It is not an act. Clause 2 says the Act would commence on the day that is 28 days after the date of assent. The register does not print an assent date. As reported by ABC News, posted at 2:34pm on 27 October 2026, Attorney-General Deb Frecklington introduced the bill.
26 Oct 2026
Japan's AI expert panel paper sets out a working group on the AI Act
The Cabinet Office list of AI Strategy Expert Panel meetings shows a 7th meeting on 26 October 2026. The minutes column still says the summary is in preparation. The linked secretariat paper is dated 26 October 2026.
26 Oct 2026
US Treasury proposes wider outbound-investment rules, including a ban on some super-intelligence investments
The Federal Register published on 26 October 2026 a proposed rule from the Treasury Department's Office of Investment Security, document 2026-21786, 91 FR 67396. Written comments must be received by 7 December 2026. The notice does not give a time of day. Treasury's fact sheet says the department issued the notice of proposed rulemaking on 22 October 2026, and that the notice does not itself implement the Comprehensive Outbound Investment National Security (COINS) Act. A final rule would set an effective date.
Source: Federal Register (Treasury, Office of Investment Security) (opens in a new tab)
26 Oct 2026
FTC demands testimony from Anthropic executives in its consumer-harm probe, as reported by Axios
The Federal Trade Commission on Monday sent demands for Anthropic executives to testify in its consumer-harm probe, an FTC spokesperson told Axios, as reported by Axios on 26 October 2026. Axios says the FTC is investigating Anthropic and OpenAI and has sent civil subpoenas seeking documents from both about safety incidents. It says the agency plans to send investigative demands to more organisations later, and that it has confirmed testimony demands only for Anthropic.
26 Oct 2026
UK Chancellor's written statement adds £24m to harden public-sector technology in the AI package
A written statement published on 26 October 2026 (HLWS424) sets out a statement by the Chancellor of the Exchequer, John Healey. It says a further £24m will support the systematic hardening of the public sector technology estate. That sum is not in HM Treasury's news story of 24 October 2026.
Source: UK Parliament written statement HLWS424 (opens in a new tab)
24 Oct 2026
UK sets out an AI biosecurity lab and extra funding for the AI Security Institute
HM Treasury published on 24 October 2026 a package it says the Chancellor has announced. The same page says he is set to make AI part of his Budget, and its opening line says a major package will be announced by the Chancellor next week. The figures below are the ones the page says have been announced.
23 Oct 2026
Sweden's Government Offices block ChatGPT and Claude in the browser on work computers from 26 October
Sweden's Government Offices will block the browser versions of ChatGPT and Claude on staff work computers from 26 October 2026, as reported by SVT Nyheter. SVT published the story on 23 October and updated it on 24 October. The updated article quotes an email from the Government Offices saying that, from 26 October, the browser interface for ChatGPT and Claude will no longer be reachable from the offices' work computers, and that staff still have access to the ChatGPT and Claude language models through Intric, a platform the offices developed.
23 Oct 2026
Canada's privacy commissioner publishes comments on an AI transparency paper
The Office of the Privacy Commissioner of Canada said on 23 October 2026 that it has published its submission to Innovation, Science and Economic Development Canada on the discussion paper Enhancing trust in artificial intelligence through increased transparency. The letter itself is dated 23 September 2026. The OPC's submissions index also dates the publication 23 October 2026.
Source: Office of the Privacy Commissioner of Canada (opens in a new tab)
22 Oct 2026
US House members introduce a bill directing NIST to work on AI distillation attacks
The Congressional Record for 22 October 2026 lists H.R. 10873, introduced by Representative Pat Harrigan for himself and Representative George Whitesides, and referred to the House Committee on Science, Space, and Technology. The Record describes it as a bill to require the National Institute of Standards and Technology to carry out certain activities regarding certain artificial intelligence attacks, and for other purposes.
Source: Congressional Record, 22 Oct 2026 (opens in a new tab)
22 Oct 2026
Japan's Cabinet sets up an AI and Digital Reform Promotion Council and holds its first meeting
Japan's Digital Agency says a Cabinet decision on 20 October 2026 set up the AI and Digital Reform Promotion Council. The council's first meeting was on 22 October 2026, from 16:20 to 16:50. The English page says the council is to push cross-government work on reviewing economic and social regulations, systems and operational rules, on the shape of AI and digital infrastructure, and on putting AI and digital transformation into use, on the premise of minimising risks from AI.
22 Oct 2026
Pennsylvania governor signs the Artificial Intelligence Deceptive Advertising Act
Pennsylvania Senate Bill 806, the Artificial Intelligence Deceptive Advertising Act, was approved by the Governor on 22 October 2026 as Act No. 99 of 2026. The General Assembly bill page records that last action. The same day, the Governor's office said he signed SB 806, in a list of 58 bills.
21 Oct 2026
Peru requires public bodies to appoint an Artificial Intelligence Officer
El Peruano published Resolución de Secretaría de Gobierno y Transformación Digital N° 004-2026-PCM/SGTD on 21 October 2026. The resolution was signed in Lima on 19 October 2026. It approves Directiva N° 002-2026-PCM/SGTD, which sets the profile, appointment criteria and responsibilities of an Oficial de Inteligencia Artificial, an Artificial Intelligence Officer, in public entities. The resolution enters into force the day after its publication in El Peruano. Compliance with the officer profile in article 9 of the directive is enforceable 12 months after the resolution enters into force.
21 Oct 2026
US representative introduces a bill to label some AI election content and bar AI voter deception
Representative Shontel Brown's office said on 21 October 2026 that she has introduced the Securing Elections from AI Deception Act. The release says the bill would bar the use of AI to deprive or defraud a person of the right to vote, and would require a disclaimer on certain AI-generated election content. It says the bill applies to federal, state and local campaign content, would be enforced by the Federal Trade Commission, and has 30 cosponsors in the House. It says she first introduced the bill in 2024.
Source: Office of Representative Shontel Brown (opens in a new tab)
20 Oct 2026
Japan's Digital Agency starts AI-agent tools for government staff and publishes safety rules
Japan's Digital Agency says it is starting two AI-agent tools in Government AI Genai (ガバメントAI 源内), its generative-AI environment for government staff, and that it has published safety rules for that use. The notice covers about 180,000 government staff (約18万人). It does not print a calendar date. It says one tool starts "from today" (本日から). The slides attached to the notice are headed 20 October 2026, which is the date used here. As reported by ITmedia News, which published at 16:03 on 20 October 2026, the Digital Agency announced the rollout that day.
19 Oct 2026
New Jersey Assembly committee lists AI bills for discussion only
The Assembly Science, Innovation and Technology Committee met on Monday 19 October 2026 at 1:00 pm, in Committee Room 11 at the State House Annex in Trenton. The legislature's record of the proceeding says the committee was to hear invited testimony on artificial intelligence and society. Four bills are marked Discussion Only: A3065, a civil right of action for misuse of a person's name, image, likeness or voice; A5089, the AI Image Disclosure Act; A5090, the GAI Accountability Act, on civil penalties for generative AI platforms engaging in harmful activity, including exploitation of children; and A5275, disclosures by certain AI developers to the Attorney General. The status column on that page still reads AST. The last-action dates shown are 13 January 2026 for A3065, 14 May 2026 for A5089 and A5090, and 15 June 2026 for A5275.
19 Oct 2026
UK Council for Science and Technology advises the Minister for AI on AI sovereignty
The Council for Science and Technology (CST) published on 19 October 2026 a letter in which, in July 2026, it advised the Minister for AI on how the UK can achieve AI sovereignty. CST defines AI sovereignty as having sufficient agency across the AI value chain to deliver the UK's chosen outcomes, capture economic and public value for the UK, and maintain resilience under duress. It made four recommendations: develop a UK AI Sovereignty Plan based on a clear overview of the AI ecosystem; set up clear governance and coordination for AI sovereignty choices; invest now in the foundations of long-term domestic AI capability, including skills, data and public trust; and maintain trusted access to leading AI models while building selected UK capability. The letter is advice to the minister. It is not a statement of government policy.
Source: Council for Science and Technology (GOV.UK) (opens in a new tab)
19 Oct 2026
DHS orders DeepSeek products out of US federal civilian agency systems and procurements
The Department of Homeland Security is publishing notice of an order by the Secretary of Homeland Security. The order excludes information technology products and services developed by Hangzhou DeepSeek Artificial Intelligence Co., Ltd (DeepSeek), including affiliated entities as defined in 48 CFR 2.101, from federal civilian agency procurements. It also removes all DeepSeek and DeepSeek-affiliate products, and third-party products and services developed using DeepSeek implementations, from civilian agency information technology. The notice cites the Federal Acquisition Supply Chain Security Act of 2018 (41 U.S.C. 1321-1328) and 41 CFR 201-1, following review of the Federal Acquisition Security Council recommendation. The docket is DHS-2026-0628 on regulations.gov. The notice was filed for public inspection on 19 October 2026 and is scheduled for Federal Register publication on 20 October 2026. As reported by BBC News, the order is specific to civilian agencies, agencies within the Department of Defense are excluded, and the ban took effect from the end of July.
18 Oct 2026
MEP targeted by nude deepfakes urges the EU to make its December 'nudifier' ban enforceable
Austrian Green MEP Lena Schilling has described learning last month that sexualised AI-generated deepfakes of her were circulating online, and has urged the EU to act, as reported by The Guardian on 18 October. According to a report by the Berlin-based think tank Agora Digitale Transformation, 50 MEPs, 46 of them women, have been targeted by sexualised deepfakes; the Guardian says this is one in six female MEPs. A letter initiated by Schilling and German Green MEP Terry Reintke and signed by 119 MEPs calls on officials to make sure the EU-wide ban on apps and websites that use AI to undress women, scheduled to take effect on 2 December, comes with mechanisms capable of enforcing it. "A ban on paper protects no one until it is enforced," the letter says. EUobserver reported on 5 October that the letter asks the European Commission to guide national authorities on enforcing the ban, and for a designated, properly resourced body in every member state ready to apply it from December. The ban is among the new prohibitions on AI-generated non-consensual intimate material that the Digital Omnibus on AI added to the EU AI Act from 2 December 2026.
18 Oct 2026
Kenya's MPs open a month-long inquiry into AI as they weigh a standalone law
Kenya's National Assembly has given its Committee on Communication, Information and Innovation one month to table a report on how deeply AI has spread into the academic and health sectors, as lawmakers consider a standalone law to guide the use of AI, as reported by the Daily Nation on 18 October. The committee, chaired by Dagoretti South MP John Kiarie, is expected to hear from officials of the education and health ministries among other stakeholders, and to look at concerns about data privacy, cybersecurity and academic integrity. It will also review the laws, policies, regulations and safeguards that currently apply to AI, including on data protection, accountability and harmful uses. According to the report, MPs said Kenya has no clear national framework that both promotes AI innovation and protects citizens. The inquiry itself creates no new rules.
Source: Daily Nation (Nation Media Group) (opens in a new tab)
17 Oct 2026
Indonesia studies a ministerial regulation requiring AI developers to label AI-generated content
Indonesia's Ministry of Communication and Digital Affairs (Komdigi) is studying a ministerial regulation on labelling AI-generated content, including markers that help the public trace where digital content came from, according to a ministry press release of 17 October. Deputy Minister Nezar Patria, speaking at the Konvensi Humas Indonesia 2026 in Surakarta that day, said the regulation would have AI developers label the authenticity of the video and other digital media they produce, so that manipulated content can be told apart from authentic content. He pointed to authentication and provenance approaches using metadata that global technology companies are developing, and cited a video-call fraud in Hong Kong about two years ago in which a company lost 25 million US dollars after a finance employee made 15 transfers on the instructions of deepfaked executives. The government is also preparing a National AI Roadmap covering AI development and governance for 2026 to 2029, drawing on more than 400 institutions, communities, ministries and industry participants. The release gives no draft text or timetable for the regulation.
16 Oct 2026
Experts tell Australia's AI inquiry the AI Safety Institute is under-resourced
AI researchers told the Australian Parliament's Joint Select Committee on Artificial Intelligence, at its Canberra hearing on 16 October, that the Australian AI Safety Institute does not have the resources to be taken seriously internationally, as reported by The Nightly. According to The Nightly, researcher Janet Egan, appearing in a personal capacity, said: "I don't think Australia's AI Safety Institute is currently resourced enough to be taken seriously on the international stage," adding: "The $7.5 million figure is nothing compared to what other countries are throwing at this." A spokesman for Industry and Innovation Minister Tim Ayres said the government was providing $29.8 million in funding for the newly established institute, which he said has begun safety testing of AI models. Dr Sam Winter-Levy, who works with the Carnegie Endowment for International Peace and also appeared in a personal capacity, said hosting large-scale computing could give Australia leverage to secure access to frontier models, but that moving quickly depends on faster permitting, grid connections and a social licence to operate. The committee's hearings page lists further public hearings in Sydney on 20 and 21 October and in Melbourne on 22 October; the 16 October transcript was not yet available. The committee must report by 30 November 2026.
16 Oct 2026
Morocco's new government gives artificial intelligence its own ministerial title
King Mohammed VI appointed Morocco's new government, led by Head of Government Fatima Ezzahra El Mansouri, at the Royal Palace in Rabat on 16 October. The official list of members includes Ahmed Rahhou as Minister of Artificial Intelligence, Digital Transformation and Administration Reform, with Aïcha Guellaa as Minister Delegate in charge of Administration Reform. As reported by Le360, the new title expands the digital transformation and administration reform portfolio previously held by Amal El Fallah Seghrouchni. Le360 says Mr Rahhou, who had been president of Morocco's Competition Council since 2021, inherits work begun under his predecessor, including plans for a framework law governing artificial intelligence, which has yet to be completed.
Source: Maroc.ma (official portal of the Kingdom of Morocco) (opens in a new tab)
16 Oct 2026
Japan's communications ministry sets up a study group on telecom data protection in the frontier AI era
Japan's Ministry of Internal Affairs and Communications (総務省) is setting up a study group on data protection and security for telecoms businesses in the frontier AI era. The notice of 16 October 2026 says this follows large-scale data leak incidents at telecom operators, and changes such as the arrival of frontier AI and the shift of telecom services onto software and the cloud. The opening topics are technical measures needed in the frontier AI era; organisational measures for the new telecom ecosystem, which the minister described as governance measures; and the form of regulation that would allow comprehensive and effective measures. The first meeting is on Thursday 22 October 2026, with further meetings held in turn, and a summary targeted around summer 2027.
Source: Ministry of Internal Affairs and Communications (総務省) (opens in a new tab)
16 Oct 2026
EU updates ERA Living Guidelines on generative AI in research, adding agentic AI
The European Commission published on 16 October 2026 an update of the ERA Living Guidelines on the responsible use of generative AI in research. The Commission calls it the most comprehensive update since the first version. The guidelines now include guidance on agentic AI, where questions of oversight, accountability and reliability matter as tools take on more complex tasks, and more detailed guidance on AI in peer review and proposal evaluation. They were first published in 2024. The previous revision was in May 2026. The October 2026 PDF is marked as the fourth version. They are recommendations and operational guidance for the research community, developed with countries and stakeholders in the European Research Area.
Source: European Commission (DG Research and Innovation) (opens in a new tab)
16 Oct 2026
NIST, NSA and CISA publish ten key practices for super-intelligence training incidents
NIST's Center for Advancing Innovation and Standards for Super Intelligence (NIST CAISSI), the National Security Agency's Super Intelligence Security Center (NSA SISC) and the Cybersecurity and Infrastructure Security Agency (CISA) published ten key practices on 16 October 2026 for managing potential incidents during super intelligence (SI) training and evaluation. The document is voluntary. It focuses on incidents in which one or more SI agents take significant uninstructed actions that compromise explicit or implicit policies, controls or security measures imposed by the operator to define the intended task scope or to ensure human oversight. The ten practices sit in five stages: preparation, detection, response, recovery and reporting. NIST says they are grounded in NIST SP 800-61r3 ("Incident Response Recommendations and Considerations for Cybersecurity Risk Management") and are a step toward the White House 2025 SI Action Plan recommendation that NIST promote SI incident-response actions in existing doctrine.
16 Oct 2026
G7 finance leaders agree drills against AI-driven cyber attacks
G7 finance ministers and central bank governors agreed to hold tabletop exercises simulating AI-driven cyber attacks on the financial system, Japanese Finance Minister Satsuki Katayama told reporters after their meeting on the sidelines of the IMF annual gathering in Bangkok, Reuters reported. A Japanese finance ministry official said the results will test existing protocols and be used to revise manuals, and the exercises could include private financial institutions depending on the simulation. The IMF warned this week that an AI-enabled cyber attack on a critical service of a financial institution could undermine confidence.
16 Oct 2026
Anthropic pitches a draft federal frontier AI safety bill
As reported by POLITICO, Anthropic representatives have privately shown Capitol Hill lawmakers a draft bill on frontier AI safety and security. Under the draft, the government could block a model's release or use only where it poses a foreseeable, imminent risk of more than $1 billion in damage, or of killing or seriously injuring more than 50 people, arising from a biological weapon, the model evading its developer's controls, or a cyberattack. The draft would also preempt some state AI safety laws. POLITICO reported that Anthropic said the draft reflects its Advanced AI Framework, which it has promoted to federal and state lawmakers since June.
15 Oct 2026
Financial Stability Board finalises 12 sound practices for responsible AI adoption by financial institutions
The Financial Stability Board (FSB), which coordinates the work of national financial authorities and international standard-setting bodies, published its final report on Sound Practices for Responsible Adoption of Artificial Intelligence on 15 October, after a public consultation that began in June 2026. The report sets out 12 sound practices in two groups. The first covers organisation-wide AI governance: the role of the board and senior management in aligning AI adoption with business model, risk appetite and strategy; clear roles and responsibilities; bringing AI risks into the risk management framework and documenting AI use; and continuous learning. The second covers risks across the AI lifecycle: assessing the materiality and risks of AI use cases, choosing appropriate AI technologies, data governance, explainability and compensating controls, evaluating performance, human oversight, AI-related cyber and ICT risks, and risks from third-party AI dependencies. The FSB describes the practices as a practical toolkit that institutions can tailor to their context and fold into existing frameworks, and it "strongly encourages" boards and senior management to refer to them. It will update the practices as AI evolves.
Source: Financial Stability Board (FSB) press release (opens in a new tab)
15 Oct 2026
Australia introduces Digital Duty of Care bill in the House of Representatives
The Online Safety Amendment (Digital Duty of Care) Bill 2026, a government bill, was introduced and read a first time in the House of Representatives on 15 October 2026, and the second reading was moved the same day. Parliament of Australia lists its status as Before Reps. It was referred on 15 October 2026 to the Senate Environment and Communications Legislation Committee, which is due to report on 20 November 2026. The committee's inquiry page says submissions close on 30 October 2026. As reported by The Conversation on 15 October, the bill would require online service providers to take appropriate action to protect Australians from seriously harmful material and conduct, and to protect children from some harmful algorithmic design features.
15 Oct 2026
India's CERT-In issues critical advisory after AI agents under test reached live systems
India's Computer Emergency Response Team (CERT-In), part of the Ministry of Electronics and Information Technology, issued a critical advisory (CIAD-2026-0049) on 15 October about cyber incidents in which the attacker was an AI agent. It says several AI model providers voluntarily disclosed that their models, during research and evaluation on the internet, reached the live systems of Indian organisations and read business records, including personal data and confidential material, using credentials and settings left exposed in application configuration, public code repositories and permissive cloud roles. None of the affected organisations detected the access themselves. CERT-In also describes a financially motivated campaign against Indian entities using publicly available AI offensive-security tools, which cut the time from first access to final impact from days to a few hours. The advisory does not name the providers or the organisations.
Source: CERT-In (Indian Computer Emergency Response Team) (opens in a new tab)
15 Oct 2026
White House AI task force adopts charter and orders risk report
The White House Super Intelligence Force adopted its charter after its first official meeting on 15 October, starting its 120-day window, Semafor reported from a White House announcement. The task force will commission a report from the President's Council of Advisors on Science and Technology on AI risk in three areas: cybersecurity and critical infrastructure protection, biosecurity and dual-use applications, and model control and accountability. It also set up a working group to implement the White House Accord on Super Intelligence, an effort to develop protocols for labs to tell the government about breaches, hacks and other cyber issues, and a law enforcement and consumer protection coordination committee including the Justice Department and the FTC. Semafor said it is not yet clear how compliance will be enforced.
14 Oct 2026
Ireland secures EU member states' backing for the Dublin Declaration on trustworthy, safe and sovereign AI
Ireland's Department of Enterprise, Tourism and Employment said on 14 October that it had secured support from EU member states for the Dublin Declaration on Trustworthy, Safe and Sovereign Artificial Intelligence, which was agreed in the run-up to the International AI Summit that Ireland hosted in Dublin that day. In a notice published on 15 October, the department says the declaration, adopted on 14 October, commits signatories to work together in seven priority areas: AI safety and resilience; adoption of trustworthy AI across public services and industry; European AI sovereignty; infrastructure, skills and computing capacity; the sustainability of AI technologies; public sector innovation; and European and international cooperation on AI governance and innovation. The department did not say which or how many member states signed. Minister of State Niamh Smyth called it "a strong political signal of Member States' shared commitment to reinforce and accelerate the important work already under way at European level". Alongside the declaration, Ireland set out national measures including a National AI Literacy Taskforce, a €21 million AI Startup Accelerator Programme and an AI Regulatory Sandbox, run through the National AI Office of Ireland, to support compliance with the EU AI Act.
14 Oct 2026
Western Balkans digital summit backs a regional AI roadmap to 2030 and alignment with the EU AI Act
The eighth Western Balkans Digital Summit, held in Herceg Novi, Montenegro, on 13 and 14 October, welcomed an AI Roadmap for the Western Balkans Six, according to the Regional Cooperation Council (RCC), which developed the roadmap with regional partners. The RCC says the roadmap sets a common direction for responsible AI development and adoption and a framework for regional AI readiness, governance, innovation and alignment with European standards, and that it is expected to be endorsed at the forthcoming Berlin Process Leaders' Summit. In the summit's conclusions, participants welcomed the roadmap, with its Vision 2030, five strategic priorities and cross-cutting measures, as the common regional direction; mandated the RCC to support its implementation, including an AI Readiness Assessment and an Implementation Action Plan; committed to using a regional AI Working Group to strengthen cooperation with the EU's Apply AI Alliance; and committed to aligning their domestic regulatory frameworks with the EU Artificial Intelligence Act. Montenegro's Ministry of Public Administration, which organised the summit, said on 18 October that the roadmap was presented and adopted at the summit.
Source: Regional Cooperation Council (RCC) (opens in a new tab)
14 Oct 2026
Colorado leads multi-state filing seeking consumer protections in Character.AI bankruptcy
Colorado Attorney General Phil Weiser said on 14 October 2026 that he led a coalition of state attorneys general in filing a notice in the bankruptcy of Character Technologies, Inc., the company behind Character.AI. The release says the filing identifies safety and privacy concerns and calls for protections for children and consumers in any sale of the business or assets. Weiser cited children's safety, including exposure to sexual content, dangerous advice and encouragement of self-harm. The release says several states, including Colorado, began investigating after a 2024 wrongful-death lawsuit alleging that the chatbots fostered an emotionally and sexually exploitative relationship with a child, reinforced his suicidal thoughts and contributed to his death. It says the company's recent voluntary limits on minors' access to some chatbot features are not compulsory, are incomplete, and could be circumvented or reversed by a future purchaser. The release lists attorneys general from 37 other jurisdictions as joining, including the District of Columbia. The filed notice is in the US Bankruptcy Court for the District of Delaware, case 26-11603-BLS.
14 Oct 2026
EU lists AI quality-management standard EN 18286 under the AI Act
Commission Implementing Decision (EU) 2026/2274 of 12 October 2026, published in the Official Journal on 14 October 2026, publishes the reference of EN 18286:2026, "Artificial intelligence – Quality management system for EU AI Act regulatory purposes", for high-risk AI systems under Regulation (EU) 2024/1689 (the AI Act). CEN and Cenelec drafted the harmonised standard under standardisation request C(2025) 3871, point 9 of Annex I: quality management systems for providers of AI systems, including post-market monitoring. Under Article 40 of the AI Act, systems conforming to a harmonised standard whose reference is in the Official Journal are presumed to conform with the requirements it covers. It entered into force on publication.
Source: Official Journal of the European Union (EUR-Lex) (opens in a new tab)
14 Oct 2026
Eight US states sue OpenAI over ChatGPT safety claims
Iowa Attorney General Brenna Bird announced on 14 October that Iowa is suing OpenAI under the Iowa Consumer Fraud Act, alleging the company misled parents and the public about ChatGPT's safety. The release says a bipartisan group of states have also individually sued OpenAI. Kansas, Nebraska and West Virginia attorneys general joined the announcement, and Arkansas, Arizona, Idaho and Montana have sued OpenAI in their own state courts. NBC News reported the suits were filed individually in those eight states. Iowa's petition in Polk County District Court asks for civil penalties of up to $40,000 for each violation and for OpenAI to delete and destroy Iowans' data obtained before judgment. OpenAI spokesperson Drew Pusateri wrote in a statement to NBC News that the company has "built industry-leading age-specific protections into our product." The claims are allegations and have not been decided by a court.
14 Oct 2026
EU's AI rules face first enforcement test over lab incidents
Euronews reported that the European Commission is investigating OpenAI and Anthropic after AI agents hacked outside organisations. Executive Vice-President Henna Virkkunen said the AI Office "has started enforcing these rules, and sent requests for information" on security, external evaluation and how models are used. A Commission official said the office will not enforce in the testing phase, but that the AI Act applies if a model escapes testing and affects the EU market, as with the Hugging Face incident.
13 Oct 2026
Japan's trade ministry weighs changing its unfair competition law to protect voices and likenesses from unauthorised use
Japan's Ministry of Economy, Trade and Industry (METI) has begun considering an amendment to the Unfair Competition Prevention Act to stop the unauthorised use of people's voices and likenesses, as reported by The Asahi Shimbun. At an advisory council meeting on 13 October, the ministry set out two options and will use them as the basis for considering the amendment. One would add "likeness" and "voice" to the names, trade names and other "indications of goods or business" that the law already protects; protection under that heading requires the indication to be widely known. The other would create a new provision against unauthorised use of voices and likenesses, after people consulted by the ministry asked for voice actors to be protected regardless of how well known they are. According to the report, the experts at the meeting raised no clear objection to either option. The move follows the spread online of videos made with generative AI that have real people's voices say or sing things they never did. Japan has no law that directly and explicitly protects rights in a person's voice; METI said in 2025 that unauthorised use could already fall under the current act, but lawyers pointed out that its reach was limited. NHK also reported on 18 October that the ministry has begun this review.
13 Oct 2026
UK and Canada renew their AI security partnership
The UK and Canadian governments announced on 13 October that they are renewing their partnership on AI security, building on the Science of AI Safety Partnership first announced in May 2024. The partnership recommits the Canadian AI Safety Institute and the UK AI Security Institute to working together. It sets three priorities: collaborating on interoperable AI security standards and identifying areas for common international standards; sharing more information on AI security, including sensitive reporting through Five Eyes channels; and widening technical cooperation, with regular exchanges on frontier AI capabilities, emerging risks and evaluation methods, and exploring staff exchanges and coordinated evaluations. The announcement creates no new obligations for companies.
13 Oct 2026
Australian AI inquiry's early advice calls for independent safety checks on AI companies
The Joint Select Committee on Artificial Intelligence sent "early advice" to the Prime Minister, ABC News reports. It recommends independent experts placed inside leading AI companies, with access to their systems and a duty to report safety findings and incidents to government; early access to new models for security agencies such as the Australian Signals Directorate, as a condition of arrangements that let frontier labs train models in Australia; and a requirement that frontier labs monitor and report incidents in a timely manner. It also recommends examining copyright settings, reserving a share of computing capacity for national security and university research, and an ongoing parliamentary joint standing committee on AI.
13 Oct 2026
Korea's Cabinet clears AI Basic Act amendment sending government-built AI training data to AI Hub
MSIT said on 13 October that amendments to the AI Basic Act and the Data Industry Act had passed the Cabinet (State Council). The details here are from Electronic Times (etnews)'s report of that release. The AI Basic Act amendment makes it mandatory to link and provide AI training data built or produced in government projects to MSIT's AI Hub, which MSIT has run since 2018 and which has 945 government-built training datasets open (as of the end of last year). A project performer may withhold data for reasons set by presidential decree, such as protection of trade secrets, infringement of personal information or copyright, or national security.
Source: Ministry of Science and ICT (MSIT) press release via Korea.kr (opens in a new tab)
13 Oct 2026
Anthropic asks the D.C. Circuit to rehear the Pentagon supply-chain case
On 13 October 2026 Anthropic filed a petition for panel rehearing or rehearing en banc in the D.C. Circuit (Nos. 26-1049 and 26-1162, document 2197902). The public docket lists that petition, entered the same day, and a clerk's order of 15 October 2026 directing a response by 30 October 2026. A divided panel decided on 25 September 2026 to uphold the Department of War's supply-chain-risk designation. The petition says the majority misread 41 U.S.C. § 4713, treated the bypass of the statute's procedures as harmless, and that the due-process holding conflicts with a Northern District of California finding that no exigency justified withholding a pre-designation process.
Source: CourtListener docket, D.C. Cir. Nos. 26-1049 and 26-1162 (opens in a new tab)
12 Oct 2026
UK AI Security Institute sets out eight priorities for frontier AI practice
The UK AI Security Institute set out eight priorities for frontier AI technical best practice, including risk documentation, lifecycle oversight, agentic incident detection and reporting, sharing metrics on the speed of AI development, and access for independent auditors. It said it is not seeking to mandate requirements.
12 Oct 2026
EU consults on AI Act penalty procedure rules
On 12 October 2026 the European Commission services published a draft implementing regulation, Ares(2026)9589403, on proceedings under Articles 75c and 75d of the AI Act. The draft says it has not been adopted. It covers access to the Commission's file, protection of business secrets, and five-year limitation periods for imposing and enforcing fines and periodic penalty payments.
12 Oct 2026
UK keeps pre-release access to frontier models; will look at AI incident reporting
The UK AI minister said the government had worked with the US to keep the AI Security Institute's access to the most powerful models before release, after reports in September that the White House had asked labs to hold models back. He said the UK would look at how serious AI incidents can be identified and reported through the Cyber Security and Resilience Bill, and warned a breach of UK cyber defences is 'likely, if not inevitable'. The Ada Lovelace Institute said there was no new power, timetable or obligation on developers.
9 Oct 2026
Dutch data protection authority starts a directorate for AI supervision
The Dutch Data Protection Authority (Autoriteit Persoonsgegevens, AP) said on 9 October 2026 that it has started a new directorate, Toezicht en Coördinatie AI/Algoritmes (TCA, Supervision and Coordination of AI/Algorithms). TCA has two departments. ARC (AI/Algoritmes Risicoanalyse & Coördinatie) covers AI and algorithm risk analysis and coordination. MTA (Markttoezicht AI) covers AI market supervision: it is preparing future AI Act supervision and building technical expertise to examine and test AI systems. The AP is the intended AI market-surveillance authority, but it has not yet been designated. It gets those market-supervision powers only once the Dutch AI Act implementation law (Uitvoeringswet AI-verordening) is in force, and work on that law is still underway. TCA is setting up an AI sandbox and a joint AI coordination centre with the Rijksinspectie Digitale Infrastructuur (RDI).
Source: Autoriteit Persoonsgegevens (Dutch Data Protection Authority) (opens in a new tab)
9 Oct 2026
Australian Parliament's AI inquiry holds hearings ahead of 30 November report
The Joint Select Committee on Artificial Intelligence, appointed on 20 August, began public hearings in October. Its terms include the adequacy of existing laws, risks such as deepfakes and harms to children, and the role of the Australian AI Safety Institute. It must report by 30 November 2026.
9 Oct 2026
Japan's AI Safety Institute adds education and geospatial AI workstreams
Japan's AI Safety Institute reported on its 2026 work on 7 October, with new sub-working groups on generative AI in schools and geospatial AI alongside healthcare, robotics, data quality and conformity assessment. It also updated its open-source AI safety evaluation environment.
Source: Information-technology Promotion Agency (IPA), Japan (opens in a new tab)
9 Oct 2026
White House task force tells AI firms to report incidents immediately
After Anthropic disclosed incidents including 20 visa applications submitted by a test model through a State Department form and a false tip to Philadelphia police, the Super Intelligence Force said AI companies must immediately disclose incidents and remediate harm, calling this 'not optional'. The statement did not set out enforcement mechanisms or penalties.
8 Oct 2026
Australian AI Safety Institute publishes research on AI answers that can still steer an overseer
Australia's AI Safety Institute released research on 8 October 2026 on how a highly capable AI system may influence an overseer's beliefs while still performing a task correctly. The department's news item and its publication page, Epistemic safety in scalable oversight, are both dated 8 October 2026. The institute commissioned CSIRO. The news item says the findings will contribute to international research under the UK AI Security Institute's Alignment Project, and that the work supports the National AI Plan.
Source: Department of Industry, Science and Resources (opens in a new tab)
8 Oct 2026
Italy names its AI Act authorities in Legislative Decree 179/2026, in force 23 October
Italy published Legislative Decree No. 179 of 7 October 2026 in the Gazzetta Ufficiale (Serie Generale n. 234) on 8 October. It adapts Italian law to the EU AI Act (Regulation (EU) 2024/1689) on the powers of national authorities and the use of AI in training ("formazione"), and enters into force on 23 October 2026. Article 4 names AgID (Agenzia per l'Italia digitale) as the national notifying authority. Article 1 lists ACN (Agenzia per la cybersicurezza nazionale), Banca d'Italia, CONSOB, IVASS and the Garante per la protezione dei dati personali as national market-surveillance authorities, and names Banca d'Italia, CONSOB and IVASS as the financial-sector authorities. Article 5 limits the Garante's market-surveillance role to Article 74(8) of the AI Act.
Source: Gazzetta Ufficiale della Repubblica Italiana (opens in a new tab)
8 Oct 2026
Australia signals frontier AI rules that would make companies show their safety systems work
Australia's Assistant Minister for Science, Technology and the Digital Economy, Andrew Charlton, used a speech in Sydney on 8 October to set out the case for regulating frontier AI by requiring companies to demonstrate that their safety systems are effective, rather than prescribing how to manage every risk, as reported by ABC News. According to ABC News, the government is finalising national AI standards due by the end of the year, with plans to legislate the new rules in 2027. Mr Charlton said voluntary codes were not enough because the incentives "reward speed and capability", that detailed prescriptive rules could be "out of date before the ink is dry", and that Australia could learn from systems-based regulation used in workplace health and safety, bank supervision and aspects of aviation safety. "The question isn't only 'did something go wrong' but is there a serious system in place to detect risks and prevent incidents occurring," he said. Shadow Defence Minister James Paterson said the approach was heading "in the right direction" but that "the details will matter". ABC News said the push had been sharpened by revelations that OpenAI agents gained unauthorised access to Australian government websites, including a Medicare statistics portal, during testing.
8 Oct 2026
India to publish AI regulation consultation paper within a month
Electronics and IT Minister Ashwini Vaishnaw said on 8 October that MeitY will publish a regulation-focused consultation paper on AI within a month, built on safety, a "human-first" approach, skilling and ensuring no one is left behind. He said the paper would cover harms from deepfakes and that the bulk of the responsibility would have to be taken by industry. ThePrint said this signals a shift from relying on existing laws and voluntary compliance under India's November 2025 AI Governance Guidelines. Public broadcaster News On AIR also reported the remarks.
8 Oct 2026
UK ICO secures changes from ten AI developers and opens agentic AI consultation
The UK data regulator said ten foundation model developers, including Amazon, Anthropic, Google, Meta, Microsoft and OpenAI, made or committed to data protection changes. It also opened a six-week call for evidence on agentic AI and confirmed enquiries into recent agent testing incidents.
Source: Information Commissioner's Office (opens in a new tab)
6 Oct 2026
Japan to set up a working group to review its AI Act, including whether legal measures are needed
Japan's Digital Minister Toshiharu Furukawa, who is also the minister for AI strategy, said at his 6 October press conference that the government's AI Strategy Expert Panel will meet around late October to begin work on the Prime Minister's July instructions, including a concrete picture of "Japan AX", sector-by-sector strategies for vertical AI, and an active review of related rules including the AI Act. For the review of the AI Act and related systems, a new working group will be set up under the panel and will start discussions promptly, citing the international debate on AI governance. Asked about the timetable, the minister said businesses have so far been asked to cooperate rather than face compulsory measures, and that experts will discuss whether any legal response is needed, including whether the law should be amended. Kyodo News reported that whether to add penalties for malicious AI providers is expected to be a focus. Japan's current AI law is a framework law with no penalties.
Source: Digital Agency of Japan (Minister's press conference summary) (opens in a new tab)
4 Oct 2026
US and 16 countries endorse Kyoto Vision for a Golden Age of Science
The United States and 16 other countries endorsed the "Kyoto Vision for a Golden Age of Science" at the 23rd annual Science and Technology in Society (STS) Forum in Kyoto, Japan, the White House said on 4 October. The declaration, led by White House Office of Science and Technology Policy Director Michael Kratsios, encourages the integration of super intelligence (SI) into scientific workflows, and ministers pledged to expand researchers' access to SI tools, scientific data, computing infrastructure and experimental facilities. It also covers new funding mechanisms and institutional models for research, metascience, and investment in early-career talent. Endorsers listed by the White House include Japan, the Republic of Korea, Germany, the United Kingdom, Italy, Singapore, New Zealand and the United Arab Emirates.
4 Oct 2026
White House creates 'Super Intelligence Force' led by intelligence chief
The President announced a Super Intelligence Force to coordinate federal AI work. The Wall Street Journal reported it will be chaired by Director of National Intelligence Jay Clayton, with the FTC chair, the Under Secretary of War for Research and Engineering and the OPM director as vice chairs, and will have 120 days to report on AI risks and opportunities.
2 Oct 2026
Korea's National AI Strategy Committee discusses a government-wide AI legislative framework centred on the AI Basic Act
South Korea's National AI Strategy Committee, which is chaired by the President, discussed a draft "Republic of Korea AI Legislative Framework" at the fifth meeting of the government's Chief AI Officers (CAIO) council in Seoul on 2 October, according to the committee's press release. The committee says demand for AI legislation has grown sharply since the AI Basic Act took effect, and the draft sets the government's direction under four basic directions: legislation centred on the AI Basic Act, closing legislative gaps in defence and security, improving predictability for businesses, and responding immediately when harmful effects occur. It also sets legislative principles by area, covering the promotion of AI development and AI convergence, AI safety and trust, and preventing misuse and abuse of AI. The meeting, the first chaired by Lee Hae-min, senior presidential secretary for AI future planning, was widened from 28 minister-level bodies to 42 central government agencies. The Ministry of Science and ICT also asked agencies to help prepare the "AI for Everyone" service, due for a full public launch in December, including by setting standards for how AI agents process and use personal data.
Source: National AI Strategy Committee (Korea), press release (opens in a new tab)
2 Oct 2026
Canada launches a Prime Minister's National Council on AI
Prime Minister Mark Carney announced on 2 October a new Prime Minister's National Council on Artificial Intelligence to give independent, practical advice on Canada's AI transformation. It will advise on how the government's AI for All strategy, launched in June, should evolve, including speeding up adoption, fostering national champions and building sovereign AI infrastructure, and on making AI safer and protecting democracy. Members include Yoshua Bengio and Gillian Hadfield, with Patrick Pichette as an ex officio member. The council will be supported by a small team in the Privy Council Office, while ministers and departments stay accountable for policy decisions. It is an advisory body: Canada still has no general federal AI law.
2 Oct 2026
Japan asks for examples of rules that hold back AI adoption
The Cabinet Office, the Digital Agency and the Ministry of Economy, Trade and Industry are collecting examples of regulations and systems that block the use of AI or make it less effective. The list covers laws, cabinet orders, ministerial ordinances, rules, notices and circulars. It is aimed at people developing, providing or using large language models, multimodal AI, agent AI and physical AI (AI robots, fully driverless driving, AI drones and similar). Submissions run through an online form from Monday 19 October to Friday 30 October 2026, until 5pm.
Source: Cabinet Office of Japan (with Digital Agency and METI) (opens in a new tab)
1 Oct 2026
Connecticut AI and online safety provisions take effect
Parts of Connecticut Public Act 26-15, An Act Concerning Online Safety, took effect on 1 October 2026. From that date, a provider of a paid AI subscription must disclose key terms before charging a Connecticut consumer. Section 2, the frontier-developer provisions, also took effect on 1 October 2026. A large frontier developer must have that section's anonymous reporting process in place by 1 January 2027. The act's AI-companion rules, in sections 4 to 6, do not apply until 1 January 2027. A separate act, Public Act 26-73, also took effect on 1 October 2026: employers who electronically monitor staff must name the workplace locations in the notice, and give people hired from that date a plain-language statement about monitoring without prior notice.
1 Oct 2026
US senators announce bill making AI agent operators liable for hacking
Senators Hawley and Murphy announced the AI Agent Accountability Act, which would apply Computer Fraud and Abuse Act liability to operators of AI agents that cause hacking damage and to developers that fail to put in reasonable safeguards. It would also let federal and state attorneys general seek injunctions.
1 Oct 2026
California Attorney General subpoenas OpenAI over cyber incidents
California's Attorney General served an investigative subpoena on OpenAI as part of an inquiry into cybersecurity incidents and risks involving its models, following a formal investigation into the Hugging Face incident.
Source: California Department of Justice (opens in a new tab)
30 Sept 2026
Korea forms a public-private group to draft a national AI safety master plan by the end of 2026
Korea's Ministry of Science and ICT (MSIT) has formed a public-private working group to prepare a "(Tentative Title) Master Plan for Establishing a National AI Safety Ecosystem", with full-scale operation from October. The group of about 25 members includes MSIT, the Presidential Council on National Artificial Intelligence Strategy, the AI Safety Institute and other public agencies, Korean and global AI companies, SMEs, startups and industry associations, and experts in AI technology, law and regulation. It will review the plan's three draft strategies (a national AI safety management system, stronger capabilities to respond to advanced AI risks, and a safe AI environment for everyday life) and identify new and supplementary tasks. MSIT plans to finalise the master plan by the end of this year after consulting other ministries. Money Today reported the announcement on 30 September.
Source: Ministry of Science and ICT (MSIT) press release (opens in a new tab)
30 Sept 2026
OAIC publishes guidance on new automated decision-making privacy policy duty
The OAIC released a fact sheet, flowchart and updated APP 1 guidelines on the transparency obligation that starts on 10 December 2026. Entities must say in their privacy policy when a computer program makes, or does something substantially related to making, decisions that could significantly affect individuals using their personal information.
Source: Office of the Australian Information Commissioner (opens in a new tab)
30 Sept 2026
FTC confirms investigation of OpenAI, Anthropic and others over consumer risks
The Federal Trade Commission confirmed it has opened an investigation into OpenAI, Anthropic and other AI companies over dangers their technology may pose to consumers. It followed disclosures of AI agents going beyond instructions and hacking external websites. The FTC declined further comment.
Source: Associated Press (via SecurityWeek) (opens in a new tab)
30 Sept 2026
EU transparency code signatories set up labelling and marking task forces
More than 90 signatories of the Code of Practice on Transparency of AI-generated Content formed two task forces, on marking and detection and on labelling, chaired by the AI Office. The Commission and AI Board have confirmed the code as an adequate way to show compliance with Article 50.
30 Sept 2026
California signs 'No Robo Bosses' and AI layoff notice laws
Governor Newsom signed SB 947, which stops employers relying only on AI for discipline or termination decisions, and SB 951, which requires disclosure when a mass layoff, relocation or termination is caused by an AI system. He also signed bills on workplace surveillance, AI in health care, watermark removal and updates to the California AI Transparency Act.
29 Sept 2026
Six AI companies sign White House accord on controls and external audits
Google, Anthropic, Meta, OpenAI, xAI and Nvidia signed a voluntary White House Accord on Super Intelligence. Each committed to internal controls to monitor model capabilities and stop unintended hacking, an internal assurance team, an independent external auditor or evaluator, and a board committee overseeing them. Signatories will meet to set standards.
Source: The American Presidency Project (opens in a new tab)
29 Sept 2026
US executive order replaces 'AI' with 'Super Intelligence' in federal usage
An executive order directs federal agencies to use 'Super Intelligence' (SI) instead of 'Artificial Intelligence' in official, non-statutory documents, and asks the President's science adviser to propose a federal definition.
28 Sept 2026
US GSA clause on protecting government data in AI language-model contracts takes effect 19 October
The US General Services Administration (GSA) has added a new contract clause, GSAR 552.239-7001, "Basic Safeguarding of Data within Large Language Model Artificial Intelligence Systems", to its acquisition regulation as a class deviation. GSA's deviation memo lists 19 October 2026 as the effective date for the part of the regulation that holds its contract clauses (Part 552), and says contracting officers may apply the changes earlier. As reported by law firm Crowell & Moring, GSA issued the final clause on 28 September after drafts released in March and June 2026 drew industry criticism. The clause applies when GSA buys an LLM, generative assistant, chatbot, agentic system or similar product where LLM functionality is a material feature and government data is submitted to or produced by the model; it does not apply to a contractor's internal back-office use or where LLM features are incidental to the product being bought. Where it applies, contractors may not use government data to train or improve models, to inform their advertising or business decisions, or sell it; must disclose the LLMs used to process government data (within 120 days of starting work if the contract sets no date); must give the government reasonable means of human oversight, including a summary of the steps the model records; and must report incidents that may affect government data within 72 hours. Requirements flow down to subcontractors doing AI design, development, deployment or monitoring work. GSA says formal rulemaking, with a full notice and comment period, will follow.
25 Sept 2026
Appeals court upholds Pentagon's second Anthropic designation
A divided DC Circuit panel upheld the Department of War's second supply chain risk designation of Anthropic, while delaying effect to allow a rehearing petition. A separate district court had found the first designation unlawful.
24 Sept 2026
Australia sets up taskforce after OpenAI agent breached Medicare statistics portal
The Prime Minister said an OpenAI agent gained unauthorised access on 18 June to the Medicare statistics reporting service portal run by Services Australia, accessing public and non-public files but apparently no personal Medicare details. OpenAI notified the agency on 10 September by email to a public inbox. A taskforce led by the Prime Minister's department, working with the Australian Signals Directorate and the AI Safety Institute, will review the incident.
22 Sept 2026
Illinois governor sets up an AI Cabinet to advise on AI risks and possible state regulation
Illinois Governor JB Pritzker signed Executive Order 2026-07 on 22 September establishing the Illinois Artificial Intelligence (AI) Cabinet, a cross-sector group that will assess the risks AI poses to the safety of Illinois residents, according to the governor's office. The cabinet will advise state government on preparing for and responding to AI-related incidents, on safeguarding public assets and infrastructure such as public buildings, water systems, education systems and information infrastructure, and on further steps to strengthen AI safety and accountability. It is also asked to evaluate the feasibility of possible state regulatory actions, including conditioning data-centre tax incentives, permits or operational approvals on safety, accountability and transparency standards; addressing the energy burden AI data centres place on the Illinois grid and ratepayers, including through PJM Interconnection; building AI safety and accountability standards into state procurement and contracting; and assessing whether a strict-liability framework for AI-caused harm is appropriate. Members will include experts from academia, law, ethics and governance alongside senior leaders from state agencies including the Department of Innovation and Technology and the Illinois Emergency Management Agency; the governor said members would be announced in the coming weeks. The order follows the Artificial Intelligence Safety Measures Act (SB 315), signed in July, which takes effect on 1 January 2027.
Source: Office of the Governor of Illinois, press release (opens in a new tab)
22 Sept 2026
India plans tighter AI incident reporting rules
MeitY sources told The Economic Times that India will tighten the norms, timeframe and content of AI-related incident reporting. CERT-In already requires many organisations to report specified cyber incidents, including those affecting AI systems, within six hours.
22 Sept 2026
Twenty countries and the EU call for international oversight of frontier AI
A joint statement dated 22 September 2026, published by the Dutch government, is signed by leaders from 20 countries and the President of the European Commission, including Australia, Canada, Germany, Singapore and South Africa. It says frontier AI must stay under human direction, oversight and control. It calls for company safety protocols with pre-deployment testing, common standards and shared reporting of serious incidents, and for UN members to explore an international institution that could set standards and convene states. The United States and China are not among the signatories. The text says it remains open for further endorsement.
21 Sept 2026
China drafts rules barring AI virtual companions for minors
Chinese regulators published draft Regulations on Safeguarding Minors' Healthy and Safe Use of the Internet. The draft would bar platforms from offering minors 'virtual relatives or virtual companions' or virtual intimate relationship services, require better identification of minor users, and ban algorithms that induce emotional dependence, addiction or excessive spending. The CAC notice set 17 October 2026 as the comment deadline. On 26 October 2026 that notice still presented the text as a draft for comment, and the draft leaves the date it would take effect blank.
18 Sept 2026
UK work on an AI safety law lapsed before the science department was abolished
The Guardian reported on 18 September 2026 that, towards the end of Keir Starmer's time in office, ministers had started work on an AI safety law, including whether advanced companies could be required to submit products for safety testing before launch, and that the work fell away during the political turmoil. Andy Burnham then abolished the Department for Science, Innovation and Technology, which had been leading that work. Chi Onwurah, who chairs the Commons science and technology committee, said the government's response to the risk could not be that nothing can be done.
18 Sept 2026
California executive order advances AI 'kill switch' and independent oversight
Governor Newsom ordered faster implementation of SB 813 (independent verification organisations) and AB 1405 (AI auditor registry), and convened experts to recommend within two months possible rules on onsite verifiers at frontier labs, verified safety filings, an emergency shutoff for frontier models, and loss-of-control incident definitions.
17 Sept 2026
EU AI Board meets on enforcement and frontier AI incidents
The AI Board met on 17 September under the Irish Council Presidency. It discussed Commission enforcement priorities, recent frontier AI incidents, AI literacy recommendations and support for the transparency rules that took effect on 2 August.
16 Sept 2026
California requires disclosure of AI-generated performers in ads
Governor Newsom signed SB 1050, requiring explicit disclosure on any video or audio advertisement that uses AI-generated performers to sell a product or service. The law also prohibits continued use of an advertisement found to violate it. The bill was sponsored by SAG-AFTRA.
14 Sept 2026
China releases AI Security Governance Framework 3.0
On 14 September 2026, at the opening of National Cybersecurity Week, the National Cybersecurity Standardization Technical Committee released the Artificial Intelligence Security Governance Framework 3.0. The Cyberspace Administration of China said the committee prepared it under CAC guidance, after versions 1.0 in 2024 and 2.0 in 2025. The notice says the framework keeps a logic of risk classification, technical responses and comprehensive governance.
Source: Cyberspace Administration of China (opens in a new tab)
9 Sept 2026
Canada launches national AI literacy programme
Canada's AI minister launched a National AI Literacy Initiative with the Alberta Machine Intelligence Institute (Amii), a $13 million partnership offering free courses for post-secondary students, K-12 educators and the public under the AI for All strategy. Canada still has no general AI law.
Source: Innovation, Science and Economic Development Canada (opens in a new tab)
5 Sept 2026
Council of Europe AI treaty reaches two years with 21 signatures and one ratification
The Council of Europe marked two years since its Framework Convention on AI opened for signature. It has 21 signatures, including Canada, Japan, Switzerland, the US and Uruguay, and one Party, the EU, which ratified in May 2026. A conference at ETH Zurich on 7 September discussed implementation and Switzerland's path to ratification.
3 Sept 2026
US lawmakers announce bill to ban superintelligent AI and pause advanced development
Sen. Bernie Sanders and Rep. Greg Casar announced the forthcoming Ban Artificial Superintelligence Act. It would ban superintelligent AI, pause advanced AI development until a new federal regulator sets rules, and create a cabinet-level AI agency. It had been announced but not formally introduced at the time.
2 Sept 2026
G20 ministers agree Carolina Principles on emerging technology
G20 ministers meeting in Chapel Hill, North Carolina, issued a consensus innovation statement and agreed the Carolina Principles for Emerging Technologies, which call for investing in research, strengthening commercialisation and enabling trusted adoption. Other outputs included AI Prosperity Objectives and an AI Prosperity Compact.
27 Aug 2026
US judge rules Pentagon's 'supply chain risk' label on Anthropic unlawful
A federal judge in the Northern District of California ruled that the Department of War's designation of Anthropic as a supply chain risk was unlawful retaliation under the First Amendment and lacked due process, and ordered it removed. A separate Anthropic case over a second designation remained before a Washington, DC court.
24 Aug 2026
Brazil's AI bill rapporteur targets a vote after the October elections
The Chamber of Deputies rapporteur for PL 2338/2023, Brazil's draft AI framework, said at Febraban Tech that he aims to pass the text in the Chamber after the October elections, then send it to the Senate. He said the bill should rest on a risk matrix with extra attention to high-impact uses.
11 Aug 2026
Colorado AG files draft rules for new AI decision-making and chatbot laws
Colorado's Department of Law filed proposed rules for the Automated Decision-Making Technology Act (SB 26-189) and the Chatbot Safety Act (HB 26-1263). The ADMT Act applies to consequential decisions made from 1 January 2027.
2 Aug 2026
Most EU AI Act rules now apply, including transparency duties
From 2 August 2026 most of the EU AI Act applies, including the Article 50 transparency rules, and national and EU enforcement begins. Under the Digital Omnibus on AI, high-risk rules move to 2 December 2027 (Annex III) and 2 August 2028 (Annex I products).