PREVIEW. This is how it will look in an inbox; it was not sent to you. Each real email carries its own unsubscribe link and Keryn's postal address. KERYN HEADLINES Daily, Friday 19 September 2025 Retrospective Covers 18 September 2025 00:00 to 19 September 2025 00:00 UTC Web: https://keryn.dev/headlines/daily/2025-09-19/ TOP DEVELOPMENTS [1] New attack on ChatGPT research agent pilfers secrets from Gmail inboxes Ars Technica: Security · Incident · https://arstechnica.com/information-technology/2025/09/new-attack-on-chatgpt-research-agent-pilfers-secrets-from-gmail-inboxes [2] PyPI invalidates tokens stolen in GhostAction supply chain attack BleepingComputer · Incident · https://bleepingcomputer.com/news/security/pypi-invalidates-tokens-stolen-in-ghostaction-supply-chain-attack [3] SilentSync RAT Delivered via Two Malicious PyPI Packages Targeting Python Developers The Hacker News · Research · https://thehackernews.com/2025/09/silentsync-rat-delivered-via-two.html SAST [4] LLM-Driven SAST-Genius: A Hybrid Static Analysis Framework for Comprehensive and Actionable Security arXiv · Research · https://arxiv.org/abs/2509.15433 [5] SecureFixAgent: A Hybrid LLM Agent for Automated Python Static Vulnerability Repair arXiv · Research · https://arxiv.org/abs/2509.16275 [6] ATLANTIS: AI-driven Threat Localization, Analysis, and Triage Intelligence System arXiv · Research · https://arxiv.org/abs/2509.14589 SCA [7] What Gets Measured Gets Managed: Mitigating Supply Chain Attacks with a Link Integrity Management System arXiv · Research · https://arxiv.org/abs/2509.14583 DAST [8] Orion: Fuzzing Workflow Automation arXiv · Research · https://arxiv.org/abs/2509.15195 [9] Evaluating the Effectiveness of Coverage-Guided Fuzzing for Testing Deep Learning Library APIs arXiv · Research · https://arxiv.org/abs/2509.14626 AGENTIC SDLC [10] 5 ways to integrate GitHub Copilot coding agent into your workflow GitHub · Analysis · https://github.blog/ai-and-ml/github-copilot/5-ways-to-integrate-github-copilot-coding-agent-into-your-workflow REMEDIATION [11] WatchGuard warns of critical vulnerability in Firebox firewalls BleepingComputer · Analysis · https://bleepingcomputer.com/news/security/watchguard-warns-of-critical-vulnerability-in-firebox-firewalls TRUST [12] CoSAI Strategic Update September 2025 CoSAI · Analysis · https://coalitionforsecureai.org/cosai-strategic-update-september-2025 [13] How CISOs Can Drive Effective AI Governance The Hacker News · Analysis · https://thehackernews.com/2025/09/how-cisos-can-drive-effective-ai.html MODELS [14] Beyond Surface Alignment: Rebuilding LLMs Safety Mechanism via Probabilistically Ablating Refusal Direction arXiv · Research · https://arxiv.org/abs/2509.15202 [15] LLM Jailbreak Detection for (Almost) Free! arXiv · Research · https://arxiv.org/abs/2509.14558 AI SECURITY [16] Toward a Unified Security Framework for AI Agents: Trust, Risk, and Liability arXiv · Research · https://arxiv.org/abs/2510.09620 [17] Model Context Protocol SEP-1488: securitySchemes in Tool Metadata for Mixed-Auth Servers Model Context Protocol · Proposal · https://github.com/modelcontextprotocol/modelcontextprotocol/issues/1488 ADVISORIES IN THIS WINDOW 5 npm advisories · 4 critical · 1 moderate 1 PyPI advisory · 1 critical All advisories: https://keryn.dev/headlines/advisories/ Titles, links and short excerpts only; every item links to its source. Preview: the unsubscribe link and Keryn's postal address are added to each email when it is sent.