Find published research by company, product, platform or technology.
Showing 161–180 of 355 dossiers
Live SQL subscriptions, a React/TypeScript SDK and scale-to-zero integration are available in four regions. Realtime is free during beta, but compute and egress still cost money, row-level security is absent, and the protocol is changing.
A follow-up security fix shows why wrapper capabilities need path restrictions as well as origin checks: tinyjs 0.48.0 now confines direct PDF writes and requires user confirmation for other locations.
The important failure is not another prompt injection. Plugin4Shell breaks the mechanism intended to guarantee that an AI-agent plugin is still the exact code a marketplace reviewed.
Laravel now has a framework-native approval flow for AI tools: approvable tools can pause an agent, surface arguments and reasons, then resume the same persisted conversation after a human decision.
GitHub's gh-aw reaches GA, making Markdown-defined agent jobs a supported Actions workflow pattern. Security boundaries, model costs and privileged triggers still matter.
Tailscale 1.104.4 is now tagged as a stable release after Tailworm's SSH and Serve findings. Audit SSH forwarding, update clients and assess headscale separately.
Stripe expands Global Payouts to 160 stablecoin markets and 100-plus fiat markets, with stablecoin-backed Issuing cards launching in 27 EU countries; the announcement is about financial-product infrastructure, not ordinary subscription checkout.
The live DeepSeek changelog and rate card still show distinct V4 Pro service after the previously announced September 14 reroute. That changes cost and model-selection assumptions.
The break is narrow but concrete: scripts and CI/CD calling the old route paths can fail, while monitoring code that expected `connections` inline must fetch it separately.
This is not one headline vulnerability fix. Gemini CLI 0.60 is a coordinated hardening pass across the plumbing that lets extensions, sandboxes, filesystem paths and MCP authentication influence an agent’s execution environment.
Cloud Run sandboxes now cover all resource types. The August 5 expansion matters for builders whose agents or automation run in batch jobs or continuously pulling workers rather than HTTP services, while the feature remains pre-GA and shares CPU and memory with the host container.
Google Ads now documents positive-only AI review summaries inside Search ads in the US, UK, Australia and India. Advertisers cannot customise the text; source, opt-out and reporting controls remain unclear.
Tailscale's new MagicRoute beta lets eligible organisations choose priority-based routes per client rather than regional grouping. It affects multi-region subnet access, service hosts and exit-node recommendations.
Vendor-run live-store testing exposes the trade-off between containment, quality and response speed. Store configuration matters, and the evaluator's commercial conflict requires care.
Firefox Enterprise 158 combines a Gecko-based browser with an on-premises management console, domain-level DLP and SIEM export. Initial sales focus on four European countries; ESR support and public pricing are not yet available.
Stripe's new Postgres destination gives SaaS teams a managed payment-data mirror, not a zero-latency replacement for every webhook.
Woodpecker's agent labels were self-reported and unsuitable for authorization. Version 3.19 adds server-held filters and patches a clone-step environment-variable leak; administrators should verify their worker policies.
DeepSeek V4.1 Flash supersedes the old Flash and Vision-Exp API lines with native multimodality, lower pricing and new architecture. Unlike those retired Flash aliases, the current DeepSeek API changelog and rate card still show V4 Pro as a distinct service.
Muse packages persistent autonomous execution, credentials, payments, app access and memory into a mainstream consumer product. A September macOS hotfix now provides an early real-world lesson: agent containment has to protect not only the cloud runtime but also the local control path into the agent.
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.