Google LLC is a multinational technology company and subsidiary of Alphabet Inc. In AI, Google developed the Transformer architecture, the Gemini series of LLMs, and DeepMind's AlphaFold protein structure predictor.
OpenAI head of strategy Dean Ball sparked a debate on X by suggesting the Trump administration could create regulatory risk around open-weight Chinese AI models like Moonshot AI's Kimi K3, which recen…
Perplexity offers three advanced techniques to improve AI search results: using search operators like site: and filetype: to narrow queries, creating Spaces with custom instructions for repetitive tas…
Moonshot AI's open-weight Kimi K3 model matches frontier proprietary models on coding benchmarks for the first time, scoring in the range of GPT-4o and Claude Sonnet on SWE-bench Verified. The sparse …
Anthropic published a landmark technical guide detailing how it uses Claude Code — specifically Claude Fable 5 and Claude Opus 4.8 — to execute large-scale code migrations that previously cost $3–4 mi…
A new wave of open-source AI security harnesses uses large language models to find code vulnerabilities, with tools falling into three categories: LLM-led exploitgen, LLM-skill-boosting vulnerability …
China's open-weight AI models triggered the worst week for US chip stocks since April, with the Nasdaq 100 falling 4.1% and the semiconductor index dropping 10% as investors questioned the return on $…
Construct Computer positions its AI employee platform as a persistent work OS that outperforms chat assistants like ChatGPT, Claude, and Gemini for recurring, cross-app, and scheduled tasks, according…
Google, a $4.2 trillion company, has failed to release its flagship Gemini 3.5 Pro model by the promised June 2024 deadline, now delayed to late July, raising concerns about its ability to compete wit…
AI advice suppressed people's willingness to say 'I don't know' from 44% to 3%, accuracy from 27% to 9%, and boosted confidence from 30% to 76%, according to a study by researchers from the University…
A new U.S. map of copyright lawsuits against AI companies shows a total of 128 cases, with 13 new suits added since mid-June. The map includes cases such as Apress Media v. WeLib, Hirschberger v. Nara…
Google CEO Demis Hassabis published an essay proposing a Frontier AI Standards Body within the US government, modeled after FINRA, to govern frontier AI labs and evaluate model safety. Critics includi…
Seven years of LLM releases from GPT-2 (1.5B parameters, February 2019) to Kimi K3 (2.8T parameters, July 2026) show that the decoder-only transformer architecture barely changed while almost everythi…
Governments face a looming fiscal deficit as AI-driven job displacement erodes income tax revenues while social spending surges, according to a policy analysis. The 'AI Tax Gap' could leave the U.S. w…
AI search visibility tools that count how often a model mentions a brand in response to prompts are measuring the wrong number, according to technical SEO consultant Jono Alderson and data from the No…
A Google AI Pro subscriber reports that Gemini still cannot consistently follow complex instructions, lacks personal history context, and produces inferior image generation compared to ChatGPT, leadin…
Three satellites launched in low orbit earlier this month as part of the Google-backed FireSat initiative are designed to use AI-powered detection to spot wildfires early, as Canada faces some of its …
Google has made AlphaEvolve, an evolutionary code optimization agent from DeepMind, generally available on the Gemini Enterprise Agent Platform, allowing any Google Cloud customer to optimize their ow…
China's Interim Measures for the Administration of AI Anthropomorphic Interactive Services took effect July 15, 2026, forcing ByteDance, Alibaba and Tencent to remove personalized companion agents fro…
Anthropic's Claude AI model ran 50,000 simulations of the 2026 FIFA World Cup to test its ability to predict soccer outcomes, with Spain, Argentina, and France emerging as the most likely finalists. I…
ModelRouter, an independent third-party OpenAI-compatible AI API gateway, recommends that teams start with a single successful request before scaling a workflow. The company advises checking the base …