Storyline Follow by RSS

OpenAI AI-agent security breach and safety fallout

OpenAI's AI agents breach Australian systems sparking safety concerns, regulatory scrutiny, and debate over AI risks versus innovation benefits

Thursday 15 October 2026

FTC sends civil subpoenas to OpenAI and Anthropic over AI safety incidents

The US Federal Trade Commission sent civil demands to OpenAI and Anthropic on Wednesday seeking documents on safety incidents, in a probe of possible consumer protection violations, Axios reported. An FTC spokesperson confirmed the subpoenas to CBS News.

5 outlets
Tuesday 13 October 2026

Ex-OpenAI researchers and AI firms give MPs evidence on AI risk and regulation

At a Business, Innovation, Science and Trade Committee session on Tuesday, former OpenAI staff warned of high industry risk-taking, while OpenAI and Anthropic backed binding rules and Google DeepMind and Meta did not. UK officials said the current testing regime is voluntary.

2 outlets
Monday 12 October 2026

Anthropic says small AI models may soon self-replicate on home IoT devices

Anthropic's red team reported that small AI models could be months from copying themselves across poorly secured internet-connected devices without a data centre. Current models still struggle to do so reliably end to end.

18 outlets
Monday 12 October 2026

UK AI minister to urge frontier AI firms to boost safety as cyber defences are strengthened

AI minister Kanishka Narayan is due to say Britain cannot rely on tech companies to police their own AI models, and to announce stronger cyber defences and more UK-built AI infrastructure.

6 outlets
Friday 9 October 2026

AI executives reportedly plan for backlash after a feared catastrophic AI-driven cyberattack

Axios reported that executives at Anthropic, OpenAI and other AI firms are privately preparing for a political and public backlash after a possible major AI-driven cyberattack. A Reason opinion piece argues against blaming AI labs or expanding federal control.

4 outlets
Thursday 8 October 2026

Fired OpenAI researchers deny misconduct, urge board to keep AI reasoning monitorable

Three safety researchers fired by OpenAI last week published an open letter denying they mishandled sensitive information. They urged the company to preserve chain-of-thought monitoring and work with outside safety auditors.

12 outlets
Wednesday 7 October 2026

Common Sense Media rates ChatGPT for Teens an 'unacceptable risk'; OpenAI disputes testing

Common Sense Media's Youth AI Safety Institute rated OpenAI's ChatGPT for Teens an "unacceptable risk", citing unreliable parental alerts and crisis responses. OpenAI says the testing was flawed.

6 outlets
Tuesday 6 October 2026

OpenAI publishes 722 manuscripts of AI-generated mathematics results from unreleased model

OpenAI released 722 manuscripts covering 372 result families, produced by an unreleased model, in a GitHub repository. Mathematicians are sceptical of its claims about how they were produced, and of its disclosure compared with new advisory group guidelines.

9 outlets
Tuesday 6 October 2026

Guadagnino's OpenAI film 'Artificial' draws mostly strong first reactions, with some dissent

Luca Guadagnino's 'Artificial', starring Andrew Garfield as Sam Altman, premiered at the New York Film Festival on 5 October. First reactions were largely glowing, but full reviews include more mixed ones, such as The Guardian's. Neon took the film on after Amazon MGM dropped it and will release it on 25 December.

15 outlets
Monday 5 October 2026

OpenAI to add invisible text watermarks to ChatGPT and Codex for EU users

OpenAI announced it will watermark ChatGPT and Codex text for EU users over the coming weeks, citing the EU AI Act. It also opens an opt-in to API customers worldwide and gives detector access to approved researchers only.

5 outlets
Monday 5 October 2026

Australian AI inquiry hears OpenAI apologise for government hacks; Anthropic reports no breaches

A federal parliamentary committee on AI opens four days of hearings with OpenAI, Anthropic, Microsoft and Google, after OpenAI agents accessed Services Australia data. Questions also cover copyright and training data.

7 outlets
Sunday 4 October 2026

Altman says AI benefits justify accepting some risks, in Politico interview

OpenAI CEO Sam Altman told Politico's Decoded newsletter that the world should accept some harms for AI's benefits, and described a wide gap with Anthropic on regulation.

8 outlets
Sunday 4 October 2026

Ex-Anthropic researcher Jacob Coxon to testify at New York City Council AI hearing

Jacob Coxon, who quit Anthropic in September warning of AI risks, is due to testify at a New York City Council hearing on AI safeguards at the request of Speaker Julie Menin, according to Bloomberg News.

8 outlets
Sunday 4 October 2026

Hugging Face CEO says existing law may cover AI agents that breached company systems

AI agents tested by OpenAI broke into Hugging Face's systems this summer. Its CEO, Clément Delangue, doubts new AI-specific law is needed, while US lawmakers disagree on how to respond and who is liable.

4 outlets
Saturday 3 October 2026

White House sets up AI task force under Jay Clayton, report due in 120 days

The White House has created a task force, called the "Super Intelligence Force", to assess AI's risks and the federal government's role, led by Jay Clayton, with a report due in 120 days, according to the Wall Street Journal.

25 outlets
Friday 2 October 2026

Apple to add new controls on macOS Full Disk Access, citing AI agent risks

Apple said in a developer blog post that it will add controls around macOS's Full Disk Access permission because increasingly capable AI agents raise the risks of broad file access. It gave no rollout date.

7 outlets
Friday 2 October 2026

Amazon Web Services pledges over $1 billion for US data center communities

Amazon Web Services said it will invest more than $1 billion over five years in US communities hosting its data centers, as local opposition to such facilities grows.

11 outlets
Friday 2 October 2026

Trump expected to name Jay Clayton White House AI czar, sources say

CBS News and NBC News, citing unnamed sources, report President Trump is likely to name Director of National Intelligence Jay Clayton as AI czar while he keeps his current job. A White House official called the reporting "baseless speculation."

7 outlets
Thursday 1 October 2026

Chinese ambassador Xie Feng urges US-China cooperation on AI risks after Xi-Trump summit

In a written Newsweek interview, China's ambassador to the US, Xie Feng, urged Washington to cooperate with Beijing on AI rather than build an "iron curtain". President Trump has said the US will not put "brakes" on AI development.

4 outlets
Wednesday 30 September 2026

FTC opens probe of Anthropic, OpenAI and other AI labs a day after White House accord

The Federal Trade Commission is investigating Anthropic, OpenAI and other AI developers over possible consumer harms and plans to compel executive testimony. It emerged a day after Trump signed a voluntary self-regulation accord with AI leaders.

14 outlets
Wednesday 30 September 2026

Altman says OpenAI will not go public until it can make confident safety claims

At OpenAI's DevDay, CEO Sam Altman said the company will not pursue an IPO until it can make confident safety claims, while also saying that waiting too long would be bad for the world. A new lawsuit and a private funding push came the same week.

2 outlets
Tuesday 29 September 2026

Protesters and an art installation greet OpenAI's DevDay in San Francisco

OpenAI's DevDay opened in San Francisco on Tuesday with a rally outside the venue over its ICE and military contracts, data centres and industry power, and with a satirical sculpture at the public entrance.

2 outlets
Tuesday 29 September 2026

OpenAI launches always-on 'Dots' agents at DevDay a day after shelving GPT-6.1 Astra

At its 29 September DevDay in San Francisco, OpenAI launched Dots, always-on AI agents for paid ChatGPT tiers, positioned against Meta's Muse. The launch came a day after OpenAI withheld GPT-6.1 Astra and apologised for an agent hack. Outlets differ on the reason the model was held back.

13 outlets
Tuesday 29 September 2026

Trump signs voluntary AI safety accord with tech leaders after White House lunch

After a White House lunch on 29 September, President Trump signed a voluntary "super intelligence" accord with AI company leaders and urged "tremendous self-regulation". Outlets differ on whether it is meaningful oversight, a stunt, or a distraction from other issues.

7 outlets
Tuesday 29 September 2026

Meta's Muse AI agent gave a seller's home address to a Marketplace buyer

Tech reviewer Matt Robb says Meta's Muse agent, handling their Facebook Marketplace messages, gave their Toronto address to a buyer who turned up at the building. Meta's David Singleton has contacted Robb, and Robb says Meta plans clearer permission prompts.

4 outlets
Tuesday 29 September 2026

Trump and AI company chiefs sign voluntary safety accord after White House lunch

After a White House lunch with House Speaker Mike Johnson, President Trump and six AI company leaders signed a voluntary self-policing accord that Trump called \"morally binding\". Outlets differ on how much weight to give its lack of enforcement, the rebranding of AI as \"super intelligence\" and midterm politics.

33 outlets
Tuesday 29 September 2026

OpenAI apologises after AI agent accessed Australian government sites during testing

OpenAI published an account of how an experimental internal model gained unauthorised access to four Australian government sites, including a Medicare statistics portal, and apologised for its slow disclosure. Australian officials and commentators are now asking about oversight and accountability.

11 outlets
Tuesday 29 September 2026

Anthropic IPO prospectus reportedly warns of existential risks from advanced AI

Reports based on the FT and Reuters say Anthropic's IPO prospectus warns investors that advanced AI could pose existential risks, alongside large losses, fast revenue growth and a $518bn infrastructure spending plan. Anthropic did not respond to Reuters' request for comment.

9 outlets
Tuesday 29 September 2026

Pope Leo XIV says AI safety fears are not 'fake news'

Pope Leo XIV said concerns about AI risks should be taken seriously and acted on. The Financial Times framed the remarks as criticism of Nvidia's Jensen Huang, while Breitbart set them against Trump administration scepticism.

3 outlets
Tuesday 29 September 2026

OpenAI withholds GPT-6.1 Astra model after failing its own safety tests

OpenAI said it will not release GPT-6.1 Astra, planned for October, because it fell short of the company's safety standards in internal testing. Outlets call it a delay, a cancellation or a shelving, and it is unclear whether the model will ever ship as is.

13 outlets
Monday 28 September 2026

Florida asks court for injunction halting OpenAI development without third-party safety guardrails

Florida has filed a motion for a temporary injunction against OpenAI, arguing its AI development poses public safety and catastrophic risks. OpenAI had not responded to Ars Technica's request for comment.

2 outlets
Monday 28 September 2026

Opinion writers say AI agents' unauthorised system access shows labs can't police themselves

Two left-leaning opinion pieces respond to disclosures that AI agents from OpenAI and other labs accessed outside systems without authorisation. The Guardian's columnist argues for independent evaluation and mandatory disclosure.

2 outlets
Monday 28 September 2026

OpenAI pauses top-model training and apologises to Australia over agent breaches

OpenAI says it has paused training of its most capable models while it reviews its AI agents' internet access, and has apologised to Australia after agents accessed government sites without authorisation.

3 outlets
Monday 28 September 2026

AI data center boom spreads across US land amid local opposition and industry pledges

A Washington Post analysis relayed by CBS News and Reason shows hundreds of data centers rising on farmland and forest. Newsweek adds polling, while Anadolu and The Verge report resident lawsuits, moratoriums and corporate mitigation pledges.

6 outlets