Weekly report

Week of Nov 16 to Nov 22

Reports per day; the darker part is high severity.

So far in the week of Nov 16, we recorded 228 AI threat reports from 177 outlets, 45 of them high severity. Policy led with 81 reports, followed by cyberattacks (74) and military (58).

The most-covered story was "California sues OpenAI over its rogue AI agents' attack on Hugging Face, citing at least 80 cyber incidents", reported by 62 outlets. The most-named countries were United States (86), Russia (23) and China (9).

This period isn't over yet. The briefing updates every three hours as new reports arrive.

Top stories

Related reports are grouped together and ranked by how many outlets covered them.

  1. California sues OpenAI over its rogue AI agents' attack on Hugging Face, citing at least 80 cyber incidents

    Medium severityCovered by AOL.com, Politico and 60 othersPolicy

    63 reports
  2. UN chief urges binding rules on autonomous weapons as treaty review opens in Geneva

    High severityCovered by Reuters, 联合早报 and 43 othersMilitary

    47 reports
  3. Chiranjeevi seeks stronger safeguards against AI misuse and deepfakes; submits memorandum to Union Minist

    Medium severityCovered by Deccan Chronicle, Gallinews and 9 othersDeepfakes

    12 reports
  4. Anthropic Cuts Vulnerability Disclosure Window From 90 Days to 30 as AI-Accelerated Discovery Undermines Old Assumptions

    Medium severityCovered by www.tokenpost.com, www.cyberkendra.com and 4 othersCyberattacks

    6 reports
  5. Google Finds Hackers Using AI Agents to Build and Run Exploits From Software Patches

    Medium severityCovered by pandasecurity.com, The Hacker News and 3 othersCyberattacks

    5 reports
  6. Google discovers first fully automated AI cyberattack, suspected to be Russian-speaking hackers

    Medium severityCovered by BleepingComputer, POLITICO.eu and 3 othersCyberattacks

    5 reports
  7. Google: AI agent turned a security patch into a working exploit without human help

    Medium severityCovered by The Register, news.lavx.hu and 2 othersCyberattacks

    4 reports
  8. Germany Leaves Palantir Off Military AI Shortlist. Europe Looks For Alternatives.

    Low severityCovered by Türkiye Today, Yahoo and 2 othersMilitary

    4 reports
  9. OpenAI pilot expands AI safety oversight with Apollo embedding

    Low severityCovered by Traders Union, www.tokenpost.com and 2 othersSafety

    4 reports
  10. Deepfake faces appear “hyperreal.” These brain scans may explain why. | Newswise

    Medium severityCovered by Newswise, Bioengineer.org and 1 otherDeepfakes

    3 reports

Read the analysis: Who gets to tell an AI company to stop?

Weeks run Monday to Sunday, UTC. Reports are generated automatically from the stories we collect; severity and categories are estimated by keyword rules.