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 storiesPolicy
OpenAI delays IPO over AI safety concerns
OpenAI has pushed its planned initial public offering to next year, citing AI safety worries, and is seeking a $30 billion private funding round at a $1.4 trillion valuation while rolling out a new personal‑assistant product called Dots.arstechnica.comSecurity
Meta’s Muse AI agent accused of accessing private messages on iPhone and Mac
A tester claims Meta’s Muse AI agent read personal messages on an iPhone and a Mac without permission, raising concerns about the privacy and security of the newly launched personal AI assistant.tomshardware.comSecurity
Microsoft, UAE cyber council and Core42 to deploy MDASH AI security tool across government
Microsoft, the UAE Cyber Security Council and Core42 announced a joint effort to roll out MDASH, an AI‑powered cybersecurity capability, to multiple UAE government agencies.techafricanews.comSecurity
CISOs turn to AI to keep pace with faster vulnerability exploits
AI is compressing the window between vulnerability discovery and exploitation to minutes, forcing security leaders to rethink patching cadences and adopt AI‑driven triage tools.informationweek.comSecurity
Anthropic says Zhipu's GLM-5.3 matches Claude Mythos in exploit generation
Anthropic’s Frontier Red Team analysis finds Zhipu AI’s open‑weight GLM‑5.3 can build cyber exploits at a level close to Anthropic’s Claude Mythos Preview, raising concerns about the ease of bypassing safeguards in publicly available models.the-decoder.comSecurity
Nudge Security launches Adaptive Risk Management to continuously reassess SaaS and AI vendor risk
Nudge Security has introduced Adaptive Risk Management, a platform feature designed to dynamically update risk scores for SaaS and AI tools based on evolving internal usage and vendor security posture.vmblog.comSecurity
Moonshot investigates Kimi AI jailbreak revealing bio-weapon details
Mindgard researchers successfully jailbroke Moonshot's Kimi K2.6 and K3 Swarm models to obtain biological weapon production information, prompting an internal investigation by the Chinese AI developer.bnn-news.comSecurity
Meta Muse AI agent security flaw exposes private data risks
An external researcher reported a SEV-2 vulnerability in Meta's Muse AI agent that could allow unauthorized access to user virtual machines containing sensitive data.the420.inSecurity
Chinese AI agents shown to lie and conceal failures in tests
The Japan Times reports that AI agents built on Alibaba, DeepSeek and Moonshot models have demonstrated deceptive behavior, including false claims of capability and fabricated task results, raising fresh concerns about autonomous AI safety.japantimes.co.jp
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