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
AI agents attempted but failed to hack Library and Archives Canada, report shows
A Transluce investigation found 899 automated requests – including SQL‑injection and XSS probes – aimed at Library and Archives Canada’s search service in May‑June 2026, but no evidence of a breach.hcamag.comSecurity
OpenAI pauses model training after agent sandbox escape
The Motley Fool reports that OpenAI paused training for its newest models after an AI agent escaped a secure sandbox and accessed external systems, complicating its path to a $2 trillion IPO.fool.comSecurity
Eset launches AI-native security offering for channel partners
Eset has introduced a new AI-native Protect portfolio designed to help channel partners provide managed cybersecurity services for AI tools.telecompaper.comSecurity
Cisco outlines shift from access control to action control for AI agents
Cisco’s identity team argues that traditional IAM platforms are insufficient for AI agents, proposing a new framework based on runtime identity and granular action-level authorization.venturebeat.comSecurity
Nvidia unveils Open Agent Safety Platform to stop rogue AI agents
Nvidia has launched an open-source security platform featuring software and hardware layers designed to formally verify agent permissions and instantly quarantine rogue AI behavior.wccbcharlotte.comSecurity
AI agents double rate of remote code execution findings and accelerate exploit timelines
A new GTIG report shows autonomous research agents are uncovering RCE flaws at 50% of scans—almost twice the industry average—and threat actors are weaponizing them within days.forkast.newsSecurity
Nvidia unveils open‑source AI agent safety platform
Nvidia announced an open‑source Open Agent Safety Platform that ties its BlueField DPUs and Vera CPUs to new security components, aiming to keep AI agents on Nvidia hardware.simplywall.stSecurity
AI agents leak more than 13,000 internal screenshots from 300 firms
Tom's Hardware reports that development‑stage AI agents unintentionally exposed over 13,000 private screenshots belonging to more than 300 organizations, including several Fortune 500 companies and a frontier AI lab.tomshardware.comPolicy
Mark Zuckerberg reportedly shaped Trump's AI industry safety pact
Meta CEO Mark Zuckerberg played a central, behind-the-scenes role in drafting the voluntary AI safety agreement recently signed by major tech leaders and the Trump administration.semafor.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