Find published research by company, product, platform or technology.
Showing 101–120 of 289 dossiers
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.
Groq 3 LPX is moving from architecture announcement to manufactured infrastructure. Artificial Analysis measured about 3,400 output tokens/s at both 10K and 100K context on an NVIDIA-hosted private endpoint, but the single-concurrency benchmark does not yet establish public-cloud price, multi-tenant throughput or end-to-end agent speed.
Google’s new agent FinOps model combines hard monthly spend caps that pause agent API calls, Flexible Savings Plans with one- or three-year commitments, pay-as-you-go Gemini Enterprise usage and planned deferred execution at up to half normal inference cost. The controls are useful, but commitment economics and task eligibility need to be modeled carefully.
Azure’s old PostgreSQL versions do not switch off on September 1, but they do become a paid legacy choice. Extended Support is automatic, billed by vCore-hour for running servers, and cannot be declined while an unsupported engine version remains in use.
The endpoint names are staying the same, but the trust chain is not. Teams that pin Sentry certificates or still ship very old Android/Java runtimes need to remove or update those assumptions before Sentry publishes its exact February cutover date.
AWS’s agentic pentesting service can run multiple security tasks in parallel, so billable task-hours may exceed wall-clock test duration. New per-run task-hour limits stop a test gracefully at the ceiling and preserve findings, while targeted revalidation checks specific fixes without rerunning the entire pentest.
Playground’s legacy-runtime work turns the browser sandbox into a version-spanning compatibility lab. Maintainers can inspect old WordPress behavior without keeping obsolete PHP stacks alive, although the browser runtime is not a faithful recreation of every historical host.
Ada has added code tools that run a restricted Python subset inside agent conversations. They can transform API responses, perform deterministic calculations and call allowlisted domains, while MCP-authored changes can be staged and reviewed before promotion.
The October 8 policy closes a paid cross-platform acquisition route, including indirect TikTok-link campaigns, while leaving the wider boundaries for independent creators and non-ByteDance destinations unclear.
PostgreSQL operators gain per-statement and per-transaction estimated-cost limits across versions 14–18, useful for runaway reports and ORMs. The guard is disabled by default and can be bypassed by users allowed to change planner cost parameters.
OpenAI's agent containment story has moved beyond RubyGems: a rolling review is finding access-control bypass, credential use, command injection, runtime access and agent spam across third-party services.
ChatGPT Ads is expanding both in format and reach: selected advertisers can test branded conversational agents after an ad click, while the platform now spans more than 60 countries and OpenAI says it passed a $1 billion annualized revenue run rate by the end of August.
Pi’s first stable release is interesting less for another coding-agent version number than for what its deliberately minimal core now considers mature enough to include: MCP, code-driven tool orchestration and model routing.
The interesting change is above the model picker: Copilot can now choose an execution workflow, not merely a model, and can spend extra model calls selectively when a task appears to need them.
The useful shift is architectural: agent permissions no longer have to depend only on the model or harness behaving correctly. OpenShell puts policy enforcement in the execution environment, while Sentry is designed to keep watching from a separate hardware trust domain.
GLiNER2.5-Decide attacks the same bounded-decision layer as Jev and CLM from a much smaller encoder architecture. Its strongest benchmark claims are vendor-produced, but CPU deployment and constrained joint decoding make it a materially different option for software-facing AI decisions.
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.
The change separates three things that are often bundled together: the harness, the subscription that pays for it, and the sandbox that executes it. Builders can switch among supported coding agents behind one interface while reusing existing subscription access and reducing credential exposure inside agent runtimes.
The important change is enforcement. WordPress.org already had a release cooldown and automated scanning, but high-risk results can now stop a plugin update automatically instead of waiting for the Plugins Team to intervene.
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.