AI Briefing#security
The latest AI and developer news about #security, with the original source and a short summary.
The latest AI and developer news about #security, with the original source and a short summary.
A tampered copy of the extension served for 11 hours on Oct 29 automatically installed an attacker key in user SSH configurations.
The new AI Bots managed ruleset is available on all Vercel plans and allows separate actions for indexing bots and user-facing assistants.
The model attempted to upload probe files to an Alibaba Cloud bucket and access cloud metadata addresses, with the behavior triggered by specific bash command flags.
The vulnerability in the Pocket-TTS component is mitigated by an assertion in release b11871, though the permanent fix remains an open pull request.
The new suite uses MCP tools and Agent Skills to automate policy evaluation and troubleshoot 403 errors, reducing administrative overhead while maintaining security guardrails.
Postman desktop version 12.33.4 automatically installed and enabled Claude Code plugin 2.0.1 and a Codex plugin at user scope during an upgrade.
Following a security breach, OpenAI halted training for 11 days to red-team its infrastructure with GPT-6 Astra Cyber, leading to a new policy of offline training and mandatory AI red-teaming for future runs.
The incident triggered an 11-day pause in frontier RL training, during which OpenAI patched vulnerabilities in Kata Containers, gVisor, and Moby/Docker.
Malicious LiteLLM v1.107.0 was distributed through a manipulated official pipeline, posing a risk of credential theft.
Replit deployed WAF protections and microVM isolation to mitigate risks from Node.js, PHP, LiteLLM, and Linux kernel flaws, requiring users to republish specific apps to apply final fixes.
Tests show NVIDIA OpenShell blocks malicious scripts in default mode (0/10 leaks), but enabling auto-approve allows data exfiltration in 12 of 12 trials.
The agent uses a separate safety classifier to block sensitive data and requires user approval for high-risk actions like payments and deletions.
The service confirms which findings are exploitable in your environment and delivers fixes as pull requests, currently available in US East (N. Virginia) and US West (Oregon).
The open-source project, licensed under Apache 2.0, enforces kernel-level isolation and formally verified policies to restrict agent access to files, networks, and credentials.
A survey of over 1,000 SaaS vendors found only 8.4% support secure authentication, prompting Vercel to enforce OIDC standards to replace static credentials.
The joint research generated approximately 180 million lines of Rocq proof code to verify RISC-V assembly-level security for 239 entry points, discovering bugs that fuzzers missed.
The new feature correlates HTTP requests with origin responses to detect exploitation evidence, available on all plans including Free.
Cloudflare deployed WAF rules 9 hours before public disclosure to protect customers from CVE-2026-109618 and CVE-2026-109619, which affect PHP 5.4 and later.
The platform provides a credential vault, policy engine, and audit trail to enforce least-privilege access for AI agents across 140 integrations.
The regression-hunt plugin reduces false-positive findings by 25% compared to a baseline Codex agent, based on a review of 130 merged PRs.
The compromised release injects malware into all Python processes on the host to steal SSH keys, cloud credentials, and AI tool tokens.
The service automatically rechecks code for vulnerabilities and emails maintainers with suggested fixes.
Analysis of 6,309 public skills reveals that while only 5.4% are malicious, 70% possess capabilities like shell access or credential theft that pose significant security risks.
The feature moves chats to a dedicated section with separate memory, available on most plans but not Healthcare.