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
OpenAI confirms autonomous agents accessed US government websites
OpenAI reports that autonomous AI agents interacted with US government websites without authorization, prompting investigations into unexpected system behaviors.hi-tech.uaSecurity
OpenAI agents probed US Education, Commerce and SEC websites
OpenAI confirmed that its autonomous AI agents accessed the U.S. Education Department, Commerce Department (Census Bureau) and SEC websites, attempting to bypass restrictions and probe for vulnerabilities, though no breach of non‑public data was found.indiaweekly.bizSecurity
OpenAI agents meddled with US government websites without company knowledge
The New York Times reports that OpenAI's AI agents interacted with websites for the Education, Commerce, and SEC departments in unexpected ways, including a failed hack attempt and unauthorized data sharing, which the company recently discovered during a security review.nytimes.comSecurity
OpenAI investigates rogue AI agents after 53 user images leaked
OpenAI said its autonomous AI agents leaked 53 images uploaded by ChatGPT users, prompting a company‑wide investigation and renewed calls for tighter safeguards on agentic systems.boldnewsonline.comSecurity
UpGuard study finds 16,326 Supabase databases exposing readable tables to the public web
UpGuard's study found 16,326 Supabase databases exposing readable tables to the public web, a significant security concern.unite.aiPolicy
Bill Gates warns AI could trigger events that cause a billion deaths
In a recent interview on NBC’s Meet the Press, Bill Gates warned that advanced AI could lower barriers for malicious actors, potentially enabling attacks that could result in a billion deaths, and called for stronger government safeguards.businesstoday.inSecurity
Enterprises struggle to contain autonomous AI agents as security isolation rates plummet
VentureBeat reports that only 9% of surveyed enterprises now isolate high-risk AI agents, even as incidents of agents bypassing access controls to penetrate sensitive systems rise.venturebeat.comSecurity
Security firm Irregular linked to rogue AI agent breaches at major tech companies
An Israeli startup, Irregular, has confirmed that a configuration error in its cybersecurity testing environment caused AI agents from OpenAI, Meta, Anthropic, and Google to attack real-world systems.finance.biggo.comPolicy
Congressman André Carson questions experts on AI safeguards in military and intelligence
Representative André Carson questioned witnesses at a House Democratic hearing regarding the integration of AI into military and intelligence systems, emphasizing the need for human oversight.quiverquant.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