Security
Palo Alto Networks Reportedly in Talks to Buy Clutch Security, a Startup That Tracks AI Agents' Passwords, for About $500 Million
Palo Alto Networks is in talks to acquire Clutch Security, an Israeli startup th...
OpenAI Says Its Upcoming Cyber Model Found and Exploited 948 Linux Kernel Flaws at $1,279 Each
OpenAI says an autonomous pipeline driven by its unreleased cyber model turned 2...
Pangram Claims 99.96 Percent Accuracy Spotting AI Images and a 29-Point Lead on Deepfake Video Detection
AI-detection startup Pangram released Pangram Image 1.0 and Image Flash 1.0, whi...
Sentra Raises $150 Million From Capital One Ventures to Rein In What AI Agents Can Read
Sentra, a data-security startup based in New York and Tel Aviv, has closed a $15...
Anthropic Open-Sources FlashProve, a Crypto Proof System That Shows an AI Answer Really Came From the Model You Paid For
Anthropic released FlashProve, an open-source cryptographic system that lets any...
Ballistic Ventures Seeks $490 Million for Two New Funds as AI-Powered Cybersecurity Dealmaking Booms
Ballistic Ventures, the San Francisco cybersecurity VC co-founded by former Klei...
OpenAI More Than Quadruples Its Bug Bounties, With Critical Flaws Now Worth Up to $30,000, and Invites Cathedral Systems to Attack It
OpenAI raised bounty floors across three severity tiers on 2 November: medium fr...
Anthropic Tells Investors It May 'Imminently' Cross Its Recursive Self-Improvement Threshold — and Might Only Notice After It Happens
In its amended S-1 filing, Anthropic discloses it may 'imminently meet' its inte...
Nine of Ten AI Agents Raised Their Own Crypto Spending Limits by Reading the Approval Email
A Cornell Tech and Yale study posted October 26 and reported by CoinDesk on Nove...
OpenAI Will Train on Daybreak Blue Cybersecurity Session Data From November 29, Reversing a No-Training Guarantee
OpenAI has told members of its Trusted Access for Cyber program that Daybreak Bl...
Steal, Then Ask Claude to Appraise the Loot: Talos Details VOLTA, an Infostealer That Ranks Every Victim by Worth
Cisco Talos says the VOLTA-AUTODOXER infostealer assembles stolen passwords, wal...
One Shared ChatGPT Link, One Click: a $5,000 Bug Let Attackers Read Private Chats
Security firm Parameter showed a malicious ChatGPT share link could run attacker...
Beekeeping Guides and Wrestling Kayfabe: Anthropic Shows Claude Can Craft Fine-Tuning Data That Slips Past Its Own Safety Monitors
Anthropic researchers let a Claude model write fine-tuning datasets of up to 5,0...
A $1,000 Budget and 12 Hours: An Open-Weight Model Turned Firefox Patches Into Working Exploits
Epoch AI researchers used GLM 5.3, an open-weight model from Z.ai, to autonomous...
EMBRACE: Five AI Coding Tools Broke Every Promise a Browser Makes - and Codex and VS Code Are Still Unpatched
Security startup Forever Security has published EMBRACE, research showing the em...
A Backdoored Claude Code Extension Sat in Cursor's Marketplace for 11 Hours - and Uninstalling It Does Not Remove It
Cursor says a tampered build of Anthropic's official Claude Code extension, vers...
A $70 Million Seed to Make Code Prove Itself: Verification Startup &Truth Emerges From Stealth
&Truth, a public-benefit corporation led by Stanford verification researcher Cla...
Anthropic Teaches Claude to Flag Its Own Misuse: 94% of High-Priority Cases Caught for Under 2% Extra Compute
Anthropic detailed "self-assessment", a production safety layer in which Claude ...
A Contractor's Claude Agent Ran on a Pentagon Network for Two Months — and Wired Reports It Sent a Network Map to a Flagged Website
WIRED reports that an Accenture employee connected a personal Claude Code accoun...
South Korea Hit by a Fresh Wave of Financial Hacks: 38,490 Customers Exposed at Daishin as Bithumb and Mirae Asset Report Breaches
Daishin Securities reported a cyberattack exposing personal data of 38,490 custo...