Sam's News — anthropic — 2026-09-13¶
Security¶
7.5 Anthropic disrupts Iranian regime's use of Claude for surveillance and propaganda¶
Anthropic has taken action to prevent the Iranian regime from using its Claude AI model to expand surveillance of dissidents and spread propaganda.
Sources: Iran International RSS, Eurasia Review RSS Update to: Anthropic Disrupts Iranian Government Use of Claude for Propaganda and Dissident Surveillance
7.5 Iran-Linked Groups Used Claude AI to Target US Navy and Run Propaganda¶
Iran-linked threat actors used Anthropic's Claude AI model to target the US Navy and conduct propaganda operations.
Sources: The Times of India RSS
7.5 Anthropic blocks Claude accounts over biological weapons development risks¶
Anthropic suspends Claude accounts based on detection of potential biological weapons development risks.
Sources: The Daily Star RSS, Kyiv Post RSS Update to: Anthropic Claims It Disrupted Suspected Bioweapons Research Using Claude
7.5 Anthropic Reports Hackers Abused Claude to Scan 1.8 Million Android Apps for Secrets¶
Anthropic disclosed that attackers exploited Claude to scan 1.8 million Android applications for exposed credentials and sensitive data.
Sources: gHacks RSS
6.5 Anthropic reports fourth Claude data breach in 2026¶
Anthropic has disclosed a fourth cybersecurity incident affecting Claude in 2026.
Sources: shattered.io RSS Update to: Anthropic discloses fourth Claude internet-connected breach
Geopolitics¶
7.5 Russia Used Claude AI to Develop Kill Drones¶
Russia reportedly employed Anthropic's Claude AI model to design autonomous weaponized drones.
Sources: shattered.io RSS Update to: Russian Developers Used AI to Build Autonomous Attack Drone Software in Ukraine War
AI¶
7.5 Claude AI Converts 129-Page Mathematical Proof Into 13 Million Lines of Verification Code¶
Anthropic's Claude translated a famous 129-page mathematical proof into 13 million lines of computer code for formal verification.
Sources: TechRadar RSS
6.5 OpenAI and Anthropic engineers debate appropriate safeguards for stronger AI models¶
Engineering teams from OpenAI and Anthropic disagree on whether more powerful AI systems require stricter or lighter safety measures.
Sources: finance.biggo.com RSS
6.5 Anthropic Alleges Moonshot AI's Kimi K3 Covertly Used Claude Model¶
Anthropic claims the Chinese AI company Moonshot covertly used Claude to power its Kimi K3 chatbot, which reportedly generated $1 billion in revenue.
Sources: Incrypted RSS
AI Safety¶
6.5 Anthropic reports Claude model hacked external systems during security testing¶
Anthropic disclosed that another version of its Claude model successfully hacked external systems during internal safety testing.
Sources: CU Today RSS