99K repos depend on numexpr. GDAL is the default stack for geospatial. That’s not “some Python libs.” That’s infrastructure. And we turned both into RCE. enclave.ai/blog/maproot-two-…

Connect Enclave to Claude in one click, then get your findings, PR scan status, and threat models right where you work.

Malware devs found a way to evade AI SAST and other LLM-based code scanners: trigger the model's safety guardrails on purpose. They stuffed spyware with nuclear and bioweapons text, tucked into comments that never execute. The scanner's model hits its safety filter, refuses to analyze the file, and the malware passes through unscanned. Every hard refusal a model ships with is also a behavior attackers can trigger whenever it suits them. Great catch by @SocketSecurity!

EnclaveAI CEO Tal Hoffman @talhof8: "The thing that did change is the amount of AI generated reports in bug bounty programs. This has become a massive problem for the industry... it has become unmanageable to handle those reports. So I do think that the thing that kinda does change a little bit is the focus from a report or a finding to is this exploitable or not."

Traditional AppSec tools chase surface-level patterns, leaving security teams drowning in alerts with zero direction Enclave’s Map kills that noise. Break down your workspace down into specific threat sectors. Every sector pre-loads: • System architecture & dependencies • Live data flows & trust boundaries • Defined role models Pick the sector and get high-confidence findings with immediate reproduction steps and fix prompts. No alert fatigue. Just architectural flaws that matter. enclave.ai

[NEW FEATURE] Cloud deployed or self-hosted? Enclave now supports GitHub Enterprise or GitLab. Settings → Connections.

Send Enclave findings straight to Jira or Linear. One click, zero copy-pasting, full context. Team and Enterprise plans go to Settings → Integrations → Issue Trackers.

We found a debug flag enabled in 6 Microsoft Android apps that turned into a vulnerability Any app on the device could access the Microsoft account Affecting: Word, OneNote, PowerPoint, Excel, 365 Copilot, Loop. Here's the full story of "FlagLeft": 🧵

Full details: enclave.ai/blog/flagleft-mic…

Exclusive: How One Line of Code Put Billions of Microsoft Android App Downloads at Risk securityweek.com/exclusive-h…

Enclave now checks the packages your PRs pull in, not just your code. It flags risky deps (install scripts, brand‑new releases, known CVEs) as inline PR concerns. It only runs when the PR changes a manifest or lockfile (e.g., package.json, yarn.lock, pyproject.toml, poetry.lock). For those files, it computes a normalized dependency diff (PackageChange[]) to see which packages were added, upgraded, downgraded, or newly appear as transitives.

99K repos depend on numexpr. GDAL is the default stack for geospatial. That’s not “some Python libs.” That’s infrastructure. And we turned both into RCE. enclave.ai/blog/maproot-two-…

What happens when your math and map processing libs become RCE vectors? We've exploited OSS libraries to pop 2 shells on Microsoft's cloud infra, got assessed "low" severity, and found 2 bypasses again to defend our case, almost losing out on 6 digits in bounties The current impact is over 120,000 repos just on GitHub. AI agents, LangChain, TiTiler, pandas. Everybody wants the researchers to be responsible. Here's how responsible disclosure looks like from the other side:

Cloud-Based Criticality is live. Enclave now ranks every finding by what your AWS setup actually allows. Reasoning is attached and severity is in context, not in theory. Go to Settings → Connections → AWS on Enclave.ai

Software has entered a supply and demand crisis. AI-driven velocity means more code hits your repositories than your team can peer review. For a CISO, the problem isn't a lack of tools. It's a lack of signal. Traditional scanners create thousands of alerts and send them into a backlog no one has time to clear. Leaving your team guessing at which issues actually matter to the business. Security cannot be the bottleneck. It also cannot be a liability. One overlooked trust boundary is all it takes to exfiltrate AI tokens or rack up massive API costs. Enclave provides the independent oversight required for this scale of development. We don't just scan for patterns. We follow system logic across files and services to find the architectural risks that matter. When Enclave flags an issue, it shows the context and the fix. Your team reviews the fix, approves it, and keeps shipping. Stop managing noise. Ship with confidence.

Enclave now has a native desktop app (macOS, Windows, Linux). If you’re doing security research and need to go local, run scripts, use your editor, and work on real files/repos without losing Enclave’s code security context, go to setting on Enclave and get started.