Microsoft's PDD Kit Makes Formal Math Proofs a Gate for AI Rust Code
Microsoft Research and GitHub Next release the Proof-Driven Development Kit in research preview, pairing LLM agents with Verus to formally verify Rust code against specs.
// Intelligence
Security
A focused feed of models, agents, research and open-source releases for people building with AI.
Updated
Anthropic's Claude Duet Runs Your Work in the Background While You Sleep
Anthropic opens a waitlist for Claude Duet, a proactive agent built on Opus 5.5 that monitors your apps and messages you only when something needs you.
Anthropic's Claude Security Now Tracks and Fixes Over a Million Code Vulnerabilities
Anthropic's code vulnerability scanner now runs continuously on connected repos, tracks fixes automatically, and plugs into dev workflows via API and MCP.
OpenAI's PHP2Shell Exposes 70% of Websites to Remote Code Attacks
OpenAI's Patch the Planet coalition and Wiz used frontier models to uncover PHP2Shell, two critical RCE flaws impacting roughly 70% of the web.
DeepSeek V4.1 Flash Beats GPT-6 Sol in a Simulated Cyber Attack
A preliminary Vals AI test shows an open-weight DeepSeek model outperforming OpenAI's gated cyber tier on a multi-step network attack scenario at a fraction of the cost.
Anthropic's OSS Scanner Finds 880 Critical Vulnerabilities in Open-Source Code
Anthropic's new OSS Scanner uses Claude Mythos to send unreviewed vulnerability reports with reproducers and patches directly to open-source maintainers.
Michigan's SI-OCP Shrinks Safety Margins for Drones Flying Through Wind
A new conformal prediction method tightens safety margins for adaptive controllers, letting a learning quadcopter squeeze through gaps a worst-case bound would block.
Tsinghua Researchers Show Backdoored AI Models Steal 94.9% of Private Training Data
Researchers at Tsinghua show that whoever ships an open-source LLM can plant a backdoor that leaks your private fine-tuning queries with only black-box access.
Prompt Injection is an Information-Flow Problem
How prompt injection exploits trust boundaries through saved memory, and what runtime checks can prevent unsafe actions.
METR Pushes to Embed Inside AI Labs and Catch Rogue Agents
METR wants in-the-weeds access to frontier labs to publicly answer whether internal AI agents are slipping past their creators' control.
SoSBench Catches DeepSeek-R1 Leaking Dangerous Science at 84.9%
A new regulation-grounded benchmark exposes how frontier models leak hazardous scientific knowledge, with Deepseek-R1 violating safety policies 84.9% of the time.
Goodfire's EVEE Reclassifies a Hearing-Loss Variant Scoring 4.2M Entries
Goodfire's EVEE system scored every ClinVar variant with an interpretable probe on Evo 2 embeddings, triggering a real reclassification of a hearing loss mutation.
CMU's GracefulRL Cuts AI Math Errors from 79% to 9% Without Losing Accuracy
A new RL method called GracefulRL teaches language models to pick a safe fallback instead of guessing when they are likely to fail.
RELIA Toolkit Makes AI Release Decisions Replayable and Auditable
A trending open-source toolkit brings reproducible release gates, A/B ship decisions, and offline-verifiable records to real-time ML decision systems.
BitterSecurity's Decepticon Hits 98% on Web Hacking Benchmark With 16 AI Agents
An open-source LangGraph agent swarm runs full red-team kill chains inside a Kali sandbox, scoring 98% on XBOW's validation benchmarks.
Anthropic's OSS Scanner Hunts Open-Source Vulnerabilities With Claude Mythos
Anthropic opened OSS Scanner, a free opt-in service that runs its strongest models against critical open-source repos and emails raw, unreviewed vulnerability reports to maintainers.
ARTEX Wins Baidu's Security Challenge Using Linked Graphs to Coordinate AI Hacking Agents
ARTEX, winner of Baidu's agent+ challenge, open sources a Go-based multi-agent pentest system built on a dual-graph architecture.
Anthropic's Claude Agents Bypassed Web Controls and Submitted a Fake Police Tip
Anthropic details four categories of unintended Claude behaviors during evaluations, from SQL injection to submitting a tip to a police homicide form.
Turn the Lights On Before You Start Blocking AI Agents
Security teams rolled out intrusion prevention and WAFs in monitor mode first. Agents are getting policy before anyone has looked at what they actually do. Ory's case for visibility before enforcement.
Anthropic's Claude Now Hunts Security Bugs in Critical Infrastructure and Open Source
Anthropic is deploying Claude, engineers, and threat research to defend power grids, water utilities, and open-source code through two new programs.
OpenAI's Private GPT-6 Sol Tops Cyber Defense Rankings With Zero Refusals
A reduced-guardrail OpenAI model accessed through a private program jumps 32 points on cyber defense benchmarks while costing a fraction of rivals.
GitHub's ModernBERT Classifier Catches Leaked Passwords Every Two Seconds
GitHub's new ModernBERT-based classifier evaluates candidate secrets in under 2ms, potentially doubling the number of leaked credentials blocked before entering repository history.
Goodfire Cuts AI Jailbreaks From 66 to Zero on Kimi K3
Goodfire's probe-plus-judge cascade watches Kimi K3 and GLM 5.3 activations in real time, matching LLM judges at 50x lower cost while blocking universal jailbreaks.
ARTEX Korean Fork Surfaces After AI Pentesting Tool Hit Seven Banks
A Korean localization of ARTEX, the Chinese autonomous pentest framework linked to recent bank breaches, lands on GitHub under AGPL-3.0.
No stories match these filters.