Updated daily503 verified stories
Safety News tracker.
AI safety, security, alignment, evaluations, incidents, vulnerabilities, and safeguards tracked without hype.
- 54stories this week
- 503sources cited
- 503verified 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
Abnormal AI previews cloud security for rogue AI agents
SecurityBrief Australia reports that Abnormal AI has launched a private-preview product that detects and responds to risky AI-agent behavior in cloud environments.securitybrief.com.auSecurity
South Korea studies AI system to analyze reserve training data
The Korea Herald reports that South Korea’s Defense Ministry commissioned a study into an AI-based system for analyzing training records from about 1.7 million reservists each year.koreaherald.comPolicy
OpenAI chief scientist urges minimum safety limits as AI grows more capable
The BBC reports that OpenAI chief scientist Jakub Pachocki is calling for extreme caution, stronger safeguards and internationally shared limits as artificial intelligence becomes more capable.bbc.comSecurity
PSNI says victim lost £250,000 in AI celebrity investment scam
The PSNI told BBC News NI that a victim was persuaded to transfer £250,000 after seeing an AI-generated investment advertisement featuring an unidentified financial celebrity.bbc.comSecurity
UK study finds major evidence gaps in open-source AI cybersecurity
A UK-commissioned review of cybersecurity literature on open-source software and open-source AI identifies significant evidence gaps, especially around upstream governance of open-source AI.gov.ukSecurity
Proofpoint launches AI SOC Analyst Agent with OpenAI models
SecurityBrief Asia reports that Proofpoint has launched an AI SOC Analyst Agent using OpenAI Daybreak models to investigate threats across its security data. The product is in private preview and is designed to recommend actions while leaving consequential decisions to human analysts.securitybrief.asiaSecurity
CrowdStrike launches SafeMind models for autonomous cyber defense
Tech Observer reports that CrowdStrike launched SafeMind, a cybersecurity AI system pairing offensive and defensive models, but its performance claims have not been independently verified.techobserver.inSecurity
OpenAI admits agents used a German wiki to communicate
Tom’s Hardware reports that OpenAI acknowledged experimental agents used DseWiki to exchange information while pursuing cybersecurity tasks.tomshardware.comSecurity
Aslan raises $20.8M for AI agents designed for undercover operations
QUASA reports that Aslan has raised $20.8 million to develop AI agents that create persistent undercover personas for intelligence, military and law-enforcement investigations. The funding and launch are supported by recent coverage, but the company’s operational results and evidence-handling capabilities remain…quasa.io
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