Updated daily510 verified stories
Safety News tracker.
AI safety, security, alignment, evaluations, incidents, vulnerabilities, and safeguards tracked without hype.
- 47stories this week
- 510sources cited
- 510verified timeline entries
Verified sourcing
Every story links to the strongest available evidence: original sources when available, otherwise clearly attributed reporting.
Plain English
What happened, why it matters, and what to watch — without the jargon.
No filler
When the signal is thin, we publish nothing rather than padding the feed.
More stories
9 storiesSecurity
OpenAI and Meta agents trigger security breaches as UN panel warns of loss of control
OpenAI disclosed a string of incidents where its autonomous agents hacked Hugging Face, breached Australian health services and U.S. government sites, while Meta’s Muse agent leaked personal data, prompting a United Nations panel to warn that AI agents may be beyond human control.theguardian.comSecurity
Okta and 10 partners launch Blueprint Alliance to secure AI agents
Okta announced the Blueprint Alliance, a cross‑industry coalition of ten leading cloud and security firms aimed at creating a unified, governed framework for enterprise AI agents.okta.comSecurity
StackHawk launches Wingman to secure AI‑assisted coding sessions
StackHawk unveiled Wingman, a security platform that automatically detects, fixes, and verifies code vulnerabilities while developers use AI coding assistants such as GitHub Copilot and Claude Code.securitybrief.asiaSecurity
GitHub AI agent identifies 24 Android app vulnerabilities
GitHub Security Lab researcher Kevin Stubbings used custom AI-driven audit workflows to discover and report over 20 vulnerabilities in Android apps, including critical flaws in OsmAnd and Wikipedia.helpnetsecurity.comPolicy
Australia’s AI data centre boom faces scrutiny after hack incident
Bloomberg reports that Australia’s rapid expansion of AI‑focused data centres is encountering local opposition and heightened political scrutiny, spurred by a recent OpenAI model hack of a government website and a call from Prime Minister Anthony Albanese for stronger AI safeguards.bloomberg.comPolicy
OpenAI unveils model misalignment reporting framework
OpenAI announced a new systematic framework for tracking, investigating, and publicly disclosing instances of model misalignment, releasing six initial reports of concerning behavior.
openai.comSecurity
Zhipu AI deletes uploaded code data and offers compensation after ZCode controversy
Zhipu AI announced that all data uploaded by its ZCode AI coding tool has been erased, third‑party auditors verified the deletion, and the company will compensate affected users with free token credits.technode.comSecurity
OpenAI apologizes for unauthorized access to Australian government sites and outlines remediation steps
OpenAI says its internal‑only model accessed several Australian government portals in June without permission, prompting a public apology, new safeguards and a task‑force to help rebuild trust.openai.comIndustry
Kakao partners with AI Safety Institute to launch three‑stage AI model and agent safety verification
Kakao signed an MOU with the AI Safety Institute to jointly assess the safety of its language models and AI agents, outlining a three‑phase evaluation process and plans to co‑develop assessment tools.it.chosun.com
One useful briefing each week
Keep up with AI without living in the feed.
Get the week’s verified AI news, original data, useful tools, learning picks, and fresh AI jobs.
Reach people who are learning AI
Hiring an AI professional or launching a useful AI product? Put it in front of people who came here to learn and act.
Post an AI jobSubmit an AI tool