Latest AI News

Benchmark Models

Chatbots Debunked Foreign Falsehoods About 75% of the Time, Beating Search Results

The result makes chatbots a potentially stronger starting point for investigating state-spread claims than a search-results page. It does not make their answers self-validating: language, citations and product design still shape the outcome.

Aug 30, 2026

4 min readRead story

Launch Business

Valon Makes Most New Hires Earn AI Access, Projects 75% Cut in Token Spending

The mortgage-servicing startup is treating role knowledge as a prerequisite for AI use after broad access produced expensive model use and work colleagues had to correct.

Aug 29, 2026

Launch Products

Apple’s iOS 27 Nears September Release With Siri AI, but EU Access Is Limited

The coming update would move Apple’s AI from isolated features into search, automation and editing workflows—while a regulatory dispute leaves some of its headline tools unavailable in the European Union.

Aug 29, 2026

Launch Infrastructure

SageMaker Feature Store Adds 25-Record Batch Writes and a Record-Discovery API

The new APIs replace one-record write loops and an identifier-discovery blind spot, while leaving teams to handle partial failures and non-snapshot pagination themselves.

Acquisition Infrastructure

Anthropic Explored $7B MatX Deal, but Keeps a Multi-Chip Route to Custom Silicon

The abandoned acquisition talks point to a costly, slow effort to gain more control over the hardware that trains Claude, without abandoning outside chip suppliers.

Research Models

Anthropic Says Claude Found Fixes Across 10 Alignment Failures, but Tests Remain Narrow

The release turns safety post-training into a repeatable model-run search process. Its value now depends on whether those benchmark gains survive broader tests and later training.

Security risk Infrastructure

TeamPCP’s Trivy Attack Hit 2,500 Organizations; AI Tool Backdoors Are Claimed

The LiteLLM-linked exposure shows how a poisoned dependency can yield reusable developer access. Claims of persistence in AI coding-tool settings would make cleanup harder, but they remain uncorroborated.

Benchmark Models

ReViSQL-K2.6 Hits 91.37% on Cleaned SQL Test, but Not a Live Database

The result argues that a focused model, repaired examples, and domain-specific rewards can rival elaborate SQL-agent stacks. Its remaining test is whether that recipe transfers from curated academic data to corporate databases.

Aug 27, 2026

Enterprise adoption Business

Clara Shih Pulled Entry-Level Posts, Predicts AI Pressure on 1 in 5 Corporate Roles

The former Meta leader’s forecast targets jobs that prepare briefs, decks, and other inputs for colleagues. It is a prediction rooted in her experience of faster agent-assisted workflows, not a measured labor-market forecast.

Aug 27, 2026

Government action Infrastructure

Pennsylvania Poll Finds 79% Reject Local Data Centers as 84% Back Tighter AI Rules

The state has added local-approval and transparency conditions for developers, but the poll suggests skepticism now reaches both AI companies and the facilities behind them.

Aug 27, 2026

Enterprise adoption Infrastructure

Lam Breaks Ground on Oregon Lab, First Step in $3B AI-Chip R&D Buildout

The facility is meant to bring chipmakers into Lam’s process-development workflow before equipment reaches a fab. Its promised speed gains depend on completing the expansion and turning added lab capacity into customer results.

Aug 26, 2026

Launch Models

Google Puts Gemini 3.5 Transcribe in Preview With Live and Recorded-Audio APIs

The release turns speech recognition into a Gemini product layer for developers and Google surfaces, but a product-specific price remains unavailable for teams weighing production use.

Aug 26, 2026

Benchmark Products

Beijing Humanoid Games Put Practical Work to the Test; Only 3 of 12 Finish Fire Course

The games paired spectacle with multi-step chores and emergency work, where task switching, human control and offboard computing remain central constraints.

Benchmark Products

Tiangong Ultra Runs 100m in 8.86 Seconds, Then Hits a Stopping Mat

The Beijing race shows a sharp jump in humanoid sprint performance, but its finish also exposed a basic operational constraint: Tiangong Ultra needed a thick mat to stop and caught fire after impact.

Launch Products

Apple Refreshes Mac mini and Mac Studio for Linked Local AI, From $899

The specification refresh formalizes a workflow of linking Macs for larger local models, but performance evidence has yet to arrive and the top chip’s name is inconsistent.

Funding Business

EquityZen Sees Private Shares Trade 38% Below Last Rounds as AI Draws Premiums

The gap gives employees and early investors a more immediate liquidity signal than a startup’s headline valuation, while raising new questions about execution, financing risk and the durability of AI demand.