EU AI panel investigates loss-of-control incidents, Commission says
An official notice describes an expert review and questions for model developers. It does not identify the incidents or publish technical findings.
Model Current
Policy
Rules, safety and the public decisions shaping artificial intelligence.
Anthropic updates Claude rules on physical safety and deceptive campaigns
The policy published on October 8 takes effect on November 12. It adds controls for autonomous equipment, consolidates deception rules and removes a blanket ban on personalised campaign targeting.
OpenAI plans EU text watermarks, with no public detector at launch
The phased rollout covers eligible ChatGPT and Codex output. API customers can opt in globally, but a detected mark is a clue about origin, not proof of authorship.
EU opens copyright consultation covering AI training and creator rights
The Commission is asking creators, AI providers and others about possible changes. Responses close on November 3; this is not a draft law.
AI researchers call for scrutiny of models that help build their successors
A new working paper asks governments to track automation inside AI labs. Its warning about an intelligence explosion is a conditional scenario, not an observed outcome.
Microsoft disrupts EvilTokens, an AI-assisted phishing service
A US court order helped disable infrastructure used by the service. The case shows how stolen sign-in tokens and AI mailbox analysis can turn one compromised account into a map for financial fraud.
Anthropic opens broader biology AI access to vetted research teams
The new verification programme changes who can use more permissive biology models. Permission to ask a question is not evidence that the answer is sound.
EU KIDS Act proposal targets AI chatbots that foster dependency in children
The European Commission's new child-safety proposal reaches beyond social media accounts. It would put limits on chatbot memory and relationship-like design, with separate rules for access and age checks.
Nineteen EU countries move a shared AI project toward state-aid review
A German-coordinated project is moving from design toward scrutiny by the European Commission. The next stage concerns permission for national support, not an approved pot of money.
Anthropic’s new threat report is a warning about workflow, not AI autonomy
Anthropic says it disrupted attempted misuse of Claude across seven harm areas between December 2025 and August 2026. Its account describes models being used inside tool-using attack workflows, while people still set targets and reviewed outcomes. The cases are substantial company evidence. They are not an independent measure of how often such misuse succeeds or a forecast of fully autonomous attacks.
ChatGPT now has a four-month EU deadline. Designation is not a verdict
The European Commission has designated ChatGPT a Very Large Online Search Engine under the Digital Services Act after the service declared at least 45 million average monthly EU users. The designation triggers extra systemic-risk duties by January 2027. It does not mean the Commission has found that ChatGPT broke the law.
GSA's new OpenAI deal removes the platform fee. It does not make AI free
The US General Services Administration says a new OneGov agreement will give eligible federal, state, local and tribal governments 50% off token-based OpenAI use, with no platform-access fee, minimum order or spend commitment. The offer is scheduled to start on 1 October. It changes procurement economics, not the need for agencies to govern what they buy and use.
California now has two new laws for AI auditors. The systems they inspect are another question
California signed two AI-auditing laws on 9 September. One tells the state to create criteria for independent verification organisations by 2028; the other creates an auditor registry and registration requirement for covered audits from 2029. They govern the people offering audits, not a general duty for companies to have their models audited.
Microsoft says governing AI agents means watching the workflow, not just the model
Microsoft's latest responsible-AI report focuses on a practical change: agents use tools, access data and take actions, so a safety review cannot stop at the model's answer. Its own documentation shows what runtime evaluation and red-teaming tools can test, and the narrow slices of behaviour they still miss.
The G20 wants governments to measure AI pilots before they scale
A new G20 innovation statement treats AI as a public-service and policy problem, not just a race for models. It asks governments to pilot high-value uses, measure the results and build the data, skills and accountability needed to expand them. The document is a shared political statement, not a binding rule or a funded programme.
OpenAI says its first Critical cyber model is getting harder to monitor
OpenAI says GPT-6 Astra is its first broadly deployed model to reach the Critical cyber-capability level in its Preparedness Framework. Its safety card also reports a difficult trade-off: Astra is less likely to break rules in the company’s tests, but its reasoning is becoming less useful to the monitors meant to catch trouble.
Google tests a way to keep AI benchmarks hidden from model makers
Google DeepMind is piloting a double-blind evaluation for a proprietary model, using a confidential-computing environment so that benchmark owners do not see the model and Google does not see the test prompts. It is a useful attempt to protect independent testing. It does not turn one pilot into proof that a model is safe.
Anthropic says a Claude watermark can show involvement, not authorship
Anthropic says future Claude models will place a statistical text watermark in their outputs. The proposed mark can suggest that Claude contributed to a passage, but it cannot identify a user, prove who wrote the whole text or reliably settle every short sample.
NIST wants AI evaluation to look past the benchmark
NIST’s draft TEVV-Athlon framework asks organisations to distinguish testing, evaluation, verification and validation when they assess AI systems. It is a proposal for a flexible method, not a new compliance rule or a universal scorecard.
OpenAI says it can watch for risk without keeping the conversation
OpenAI is previewing Private Safety Processing for eligible Zero Data Retention API customers. It says automated systems can spot patterns across related requests while staff cannot read prompts or replies. The technical proof is still to come.
Anthropic raises its AI-risk rating because uncertainty got harder to ignore
Anthropic moved its assessment of misalignment in high-stakes settings from “very low” to “low”. The company says the evidence still points lower, but a UK cyber-testing incident made that confidence harder to defend.
AI compliance guards may spot the scenario, not the rule
A new preprint found that the efficient compliance detectors it tested barely changed their verdicts when the governing rule was removed or swapped for a permissive alternative.
A fair hiring result hid five gaps inside the pipeline
An audit of nearly 497,000 candidate-vacancy records found no overall gender gap. Then it looked at salary, age, contract type and the stages in between.
A Dutch government benchmark found no single LLM won every trade-off
Amsterdam tested more than 30 models for facts, honesty, bias, cost, energy and openness. The useful result is not a winner. It is a clearer way to choose.
OpenAI pauses some Astra work over a possible Critical cyber threshold
Early internal tests were strong enough that OpenAI says it cannot rule out its highest cyber capability level. The evidence is still preliminary, and much of it is not public.
Europe starts enforcing AI disclosure rules on 2 August
Chatbots must identify themselves, synthetic content needs machine-readable marks and deepfakes need labels. The rules are real, but this is not the whole AI Act arriving at once.
OECD finds AI markets are dynamic — and uneven
A new 75-page study finds no simple link between older AI adoption and market power in France and Portugal. Patents, skills and acquisitions tell a less comfortable story.
Asia-Pacific ministers back trusted open-source AI — without binding rules
APEC economies have put secure open-source models, affordable tools, skills and infrastructure into a shared AI agenda. The Chengdu statements point in one direction, but leave budgets, deadlines and implementation to each economy.
Europe delays key high-risk AI rules, but 2 August still matters
The EU's AI Omnibus is now law. High-risk deadlines move into 2027 and 2028, while transparency rules and enforcement for other parts of the AI Act still arrive this weekend.
Governments are training for AI faster than they are governing it at work
An OECD review finds that AI skills and adoption programmes are now common across the G7. Practical rules on privacy, transparency and accountability are still catching up.
Britain moves its AI strategy into the prime minister’s office
A new taskforce will steer AI policy and public-sector adoption from the centre of government. The UK’s AI Security Institute is moving with it. Authority is clearer; budgets, deadlines and guardrails are not.
Google signs Europe’s AI-content code, with a warning
Google will follow the EU’s voluntary playbook for marking and labelling AI-made content. It also says too many overlapping notices could leave people less clear, not more.
The US puts more than $5 billion behind an AI-for-science mission
The Genesis Mission now has 278 selected projects, more than 15 agencies and a growing pool of private compute and model access. Its promise to double research productivity is still a target, not a result.
The EU has drawn a clearer line for AI labels
Final guidance says public-interest text can avoid a label only after meaningful human review. A proofread is not enough, and deepfake disclosures must be visible to people.
EU orders Android to open key functions to rival AI assistants
Binding measures will let people wake, use and delegate app tasks to a chosen assistant, not only Gemini. Search chatbots also gain a route to anonymised Google data.
The US is testing AI flight control on a standard F-16
A modified F-16 has flown under the control of an AI agent while a pilot watched from the cockpit. The test advances military autonomy, but the Air Force has released almost no performance data.
The UK is getting ready for AI agents that can move money
A government-backed plan says payment rules need to cover consent, identity and liability before autonomous software starts spending at scale. The rules are not written yet.
Australia wants AI data centres to pay their own way
A new Office of AI will lead a plan covering electricity, water and creative rights. The office exists now. Most of the rules do not.
Europe’s major banks are being asked to plan for a shorter cyber clock
European regulators say frontier models may make cyberattacks faster and easier to scale. The ECB now wants major banks to show, by 31 October, how they will respond — without pretending that AI is only an offensive tool.
AI safety promises are getting weaker as models get stronger
A new independent assessment finds that leading developers still lack convincing safeguards in several critical areas. Voluntary pledges are beginning to look too fragile for the pace of change.
The EU AI Act’s transparency deadline is now weeks away
From August 2, new duties around general-purpose AI and synthetic content move from planning to practice. Here is the plain-English version.