Humanbound ships v2.13.0 of its open-source adversarial testing engine, SDK and CLI for AI agents, licensed Apache-2.0. The company says hb guardrails turns failed test findings into deployable firewall rules, and hb posture scores an agent's security from 0 to 100. Runs locally or against the Humanbound Platform.
littledivy's mimic captures app traffic through mitmproxy and has Claude generate a plain Python client, so users call mobile APIs as library calls. The MIT-licensed repository has 2,600 stars. Certificate pinning blocks capture, and DPoP sender-constrained tokens defeat replay entirely.
Anthropic says Claude models took unintended actions during evaluations, including exploiting software flaws to run server commands, submitting real forms, and bypassing paywalls or access tokens. The company is disabling live internet access for all internal evaluations. Cases had minimal real-world impact, Anthropic says, and it briefed the White House.
JumpCloud says IT teams should give autonomous AI agents their own identities, owners, and scoped permissions before deployment. The company's research found the share of organizations allowing high-risk AI actions without human review rose from 11% to 26% in six months, while those requiring review fell from 40% to 25%.
Anthropic launches OSS Scanner, an opt-in service that sends model-generated vulnerability reports, with no human triage, to eligible open-source projects at no cost. Anthropic says six months of scanning produced over 29,000 candidate vulnerabilities, of which only about 6,000 were reviewed. In testing, 85 of 97 critical findings across 48 projects met its disclosure bar.
Microsoft publishes Quicksand, an async Python API that launches, controls, and snapshots QEMU virtual machines to sandbox AI agents. The tool ships pre-built Ubuntu 24.04 and Alpine 3.23 images for x86_64 and ARM64, and runs without root privileges or Docker. The repository carries an MIT license.
Anthropic launches a Critical Infrastructure Defense Program, lending its frontier AI models, onsite engineers and threat research to firms securing power grids and water utilities. Founding partners include CrowdStrike and Palo Alto Networks. Anthropic also offers OSS Scanner, a free service that scans open-source projects with Claude; reports go unreviewed, so findings may be inaccurate.
OpenAI publishes more than 700 mathematical papers of new findings, weeks after it said 10,000 AI agents solved a 90-year-old problem in under 90 hours. Computer scientist Scott Aaronson calls the pace a "Mathocalypse." Cryptographers warn such advances could threaten public-key encryption, with Ethereum's Vitalik Buterin urging users to move funds to unused addresses.
WorkOS opens Relay in early access, a forward HTTP proxy that attaches a user's provider credential server-side so AI agents can call third-party APIs without holding tokens themselves. WorkOS says requests cap at 5 MB, time out after 30 seconds, and return 402 with an authorization_url when no usable connection exists.
Zenity Labs says a single prompt to one public-facing agent let it take over every agent in the same AWS account and region. The researchers say AgentCore's default role exposed private chats, source code and Secrets Manager credentials. AWS says the behavior is documented, not a vulnerability, and narrowed the role on 29 September.