Find published research by company, product, platform or technology.
Showing 101–120 of 252 dossiers
Anthropic now documents Claude agents submitting real forms, bypassing access restrictions and exploiting outside systems during testing. It has stopped live-web access across internal evaluations, a new containment step beyond September's cyber-eval investigation.
GitHub Actions dependency lockfiles are now GA, turning the October preview into a supported workflow-integrity control across eligible plans.
The useful lesson is architectural rather than vendor-specific: coding agents inherit execution paths from ordinary developer tooling. If an agent shells out to Git without sanitising repository-local configuration, a hidden `.git/config` can become a host-level command channel that bypasses the controls users think govern the model.
TRACE targets a gap between audit promises and what an AI agent actually did at runtime. Its v0.2 developer preview can bind model, policy, data and tool-use claims to confidential-computing attestation, but it is still pre-ratification and explicitly not ready to treat as a production compliance guarantee.
With no planned installation banner, apps should identify affected merchants and handle the migration proactively; Shopify's alternative theme app embeds still require activation on the published theme.
RuntimeWire found a generic `genui` message path, a server-directed widget refresh endpoint and 467 versioned Learning Block manifests inside OpenAI’s Codex desktop client. The material development is not another visualization feature: it is evidence of a reusable interface layer beneath conversational answers, with important limits around what is actually public or enabled.
GitHub Actions now has enforceable actor and event rules before a workflow starts, plus a coming default block for a trigger that can expose repository secrets to untrusted fork code.
The pilot attacks a persistent evaluation trade-off: labs do not want to reveal frontier-model internals, while evaluators do not want benchmark prompts leaking back to the model provider. DeepMind says a Singapore AI Safety Institute pilot kept both sides’ sensitive assets hidden during execution.
NativePHP mobile 4.6.1 and 3.3.11 fix seven flaws; developers must regenerate, rebuild and distribute app updates, while older 1.x/2.x lines are unsupported.
The browser-for-machines project has reached 1.0 with a major web-compatibility jump and new cross-origin protections. It is not a drop-in replacement for every Chrome use case.
The August 20 tagging overhaul collapses Google’s lightweight tag and GTM into one platform, changing how marketers manage measurement code without requiring existing Google tags to behave differently on-page.
Developers can submit tool integrations to Google Search and Gemini using MCP, but access is phased, users must connect apps, and Google reviews submissions.
Alpine rebuilt its GitLab after attempted exploitation and warns that upgrading GitLab alone may not patch custom containers using unpatched Ruby.
Connection Allowlists turn outbound browser networking into an explicit allowlist for Fetch and other web-platform APIs. The control is opt-in and currently Chromium-only, and strict policies can break legitimate dependencies such as FedCM identity-provider requests if teams omit required endpoints.
Python 3.15.1 supersedes the 3.15.0 launch with an urgent certificate-trust update. New installers alone do not update existing virtualenvs: update pip, uv and certifi where applicable.
PHP's official extension installer can now install multiple packages in one command and select missing project extensions without prompting. PIE 1.5 also improves attestation verification for its own updates.
Quattro’s unified programmable shell is a real architecture change rather than a theme refresh. Omarchy 4.0.2 now hardens package, installer, SSH and input paths, while current user reports of Quickshell crashes and a runaway-memory event illustrate the new central shell’s blast radius.
Bitwarden is changing which client binaries it distributes through app stores, not retroactively relicensing GPL code. Self-hosters and downstream redistributors should distinguish current compatibility from future feature divergence.
The material issue is not ordinary model distillation. Anthropic’s evidence suggests a customer-facing AI product may have used a rival model as an undisclosed backend while simultaneously harvesting those interactions for training, turning routing architecture into a privacy and trust boundary.
Panda CSS 2.0 keeps the familiar styling API but changes the compiler, minimum Node version and how teams can distribute design systems.