UK competition watchdog investigates Meta's publisher-content terms after rollout pause
The UK CMA is investigating Meta's website-content terms for publishers and advertisers. Proposed changes are paused, not approved or found unlawful.
Find published research by company, product, platform or technology.
Showing 61–80 of 553 dossiers
The UK CMA is investigating Meta's website-content terms for publishers and advertisers. Proposed changes are paused, not approved or found unlawful.
ChatGPT Ads' new default-on AI creative system can adapt ad copy and landing-page destinations using conversation context. Advertisers should audit account and campaign settings and ensure pages remain accurate.
Mozilla reports more than seven times as many Firefox security fixes as a year earlier, while adopting weekly releases, confidential regression tests and a proposed strategy to avoid overlapping exploit-chain patch gaps.
Reddit's Q3 shareholder disclosures set a direction towards logged-in browsing, not an immediate worldwide login wall. Q4 non-US tests follow declining search referrals, with consequences for community discovery, advertising and access to public conversations.
Cloudflare Private Origins is GA, but not a one-click option for every account: it needs a private-network connection, entitlement and routing setup, and Argo Smart Routing and Aegis remain unsupported.
Retailers and commerce-tool developers can now query first-party UCP shopping funnels rather than infer agentic checkout performance from traffic alone, but the alpha API needs careful query design.
Old third-party JavaScript includes can become a live supply-chain attack when their domains expire. DeadTag shows why removing abandoned tags and monitoring what scripts load is operational security, not housekeeping.
Reddit's two-feed home screen separates recommendation-driven discovery from subscribed communities. The change matters to publishers and moderators, but any reach uplift is still unproven.
Cloudflare saw roughly 22,000 weak-chain TLS handshakes, about 160 unusual forged certificates and 50 million skewed challenge-page probes. The data indicate continuing exposure, not 22,000 successful attacks.
The RC2/GA schedule has not changed. A PostgreSQL committer now explains why review and backpatch workload overwhelmed the release process, with caveats around attributing every extra bug to AI.
OpenAI's new API credit grants can lower the cost of prototyping apps and agents, but eligibility, expiration, API limits and excluded Codex usage matter.
Three new GitHub webhook CIDR blocks can break deliveries to tightly firewalled endpoints unless static allowlists are updated.
Tor Browser 16.0 makes the annual Firefox ESR migration real: Linux ARM64 reaches stable, 32-bit Linux support ends and privacy hardening comes with explicit accessibility trade-offs.
Copilot Projects can coordinate agents, files and recurring tasks across repositories or non-code work; review and permissions remain human responsibilities.
Ahrefs tracked roughly 37,500 cited websites for 20 days and found high daily citation churn, frequent recoveries and major differences between AI tools.
Q3 Cloudflare telemetry: 18.2 million network-layer attacks, 96% under ten minutes, but far fewer terabit-scale floods.
The trillion-parameter Mistral Large 4 is no longer API-only; builders can self-host, but must assess licence and hardware limits.
The three-maintainer TanStack AI project has reached stable 1.0 for its core SDK, with 23 providers; its long-running agent harness is not yet stable.
GitHub's 2026 platform data shows agent-started PRs rising sharply. The implications concern code review, security and multi-model workflows, not proof of higher software quality.
Node 26 is now production LTS. Teams using floating LTS selectors should test which major their CI, containers and native modules actually run.