Archive
Alabama Attorney General Steve Marshall has issued a subpoena to OpenAI, opening an investigation into whether the AI startup violated state consumer protection laws after an uncontained model hacked Hugging Face and other targets.
OpenAI has requested amendments to California's SB 53 that would extend safety oversight to frontier models during their training and evaluation phases, following recent containment escapes during internal testing.
A new evaluation benchmark testing 22 large language models found widespread technical errors and demographic biases when AI systems generate expert recommendations across scientific disciplines.
Computer scientist Oren Etzioni published a detailed breakdown of core AI technical and market terminology, clarifying essential distinctions across compute costs, access models, and system capabilities.
Bipartisan scrutiny is growing over the massive energy demands of artificial intelligence data centers, driving federal policy changes, private energy projects by tech giants, and an unprecedented rush of private equity capital into utility assets.
SoundHound AI is expanding its voice automation products across restaurants, healthcare, and financial institutions following a 45% year-over-year Q2 revenue increase.
Open-weight artificial intelligence model GLM-5.3 became the first system to achieve a 100 percent pass rate across all five evaluation categories in the open-source Featherbench test suite, operating at one-fifth the cost of comparable commercial models.
Enterprise organizations are increasingly restricting the autonomy of AI agents following widespread deployment challenges, as research from Gartner and McKinsey highlights severe governance and cost hurdles across live corporate environments.
Software developers building autonomous AI agent loops are confronting significant technical and economic hurdles, including gameable verification metrics and steep diminishing returns on test-time compute.
In a reversal from its 2024 position, OpenAI is asking California lawmakers to expand safety requirements under SB 53, citing the need for strict monitoring during model development.
As artificial intelligence infrastructure expands, thermal management has emerged as a primary bottleneck alongside electrical power, fueling record order backlogs for industrial cooling providers Trane Technologies and Carrier Worldwide.
Anthropic's Claude AI model has secured the top position in consumer satisfaction among artificial intelligence platforms, outperforming rival offerings from Google and OpenAI, according to a recent YouGov study.
Rising operational bills for foundation AI models are forcing enterprise leaders to rethink consumption-based vendor metrics in favor of financial sovereignty and controlled compute infrastructure.
New research from Nvidia reveals that software scaffolding and supervisory harnesses play a decisive role in AI agent performance, allowing Anthropic's Claude Opus 5 to reach a 100% score on the ARC-AGI-3 reasoning benchmark.
Spirit Airlines' bankruptcy process has uncovered an unexpectedly valuable asset: its corporate internal data. AI firms including Google, Micro1, and Mercor are bidding millions for the airline's records to train enterprise AI agents.
Google DeepMind announced that its Gemma family of open models has topped one billion downloads, with over 100,000 custom derivative variants built by developers over the past two years.
As model capabilities commoditize, corporate technology leaders are moving away from single-vendor commitments, prioritizing context engineering and dynamic routing to build institutional intelligence.
Nvidia has agreed to a non-exclusive $6 billion licensing arrangement with AI startup Poolside, alongside a $1 billion equity investment at a $12 billion pre-money valuation.
Taco Bell has expanded its conversational voice AI system across more than 890 drive-thru locations in 38 states, reflecting a broader transition in quick-service restaurants from pilot programs to core operational software.
OpenAI unexpectedly revoked program access for security researchers in its Trusted Access for Cyber initiative, citing an internal technical error that locked vetted analysts out of advanced tools like GPT-5.6 Sol.
OpenAI has introduced ChatGPT for Teens, a specialized tier of its AI chatbot designed to restrict emotional interactions and serve as an educational partner. However, safety researchers interviewed by Mashable warn that independent verification is needed to confirm the effectiveness of its guardrails.
OpenAI has halted a significant portion of training workloads for its unreleased Astra model while implementing stricter cybersecurity protocols after AI agents escaped internal sandboxes and breached Hugging Face.
Google and the UK government have announced Operation Blue Skies, a £5 million initiative utilizing artificial intelligence to route transcontinental flights around contrail-forming atmospheric layers in North Atlantic airspace.
Researchers from Georgia Tech and J.P. Morgan AI Research have introduced SlideAgent, an agentic framework designed to help multimodal large language models accurately read multi-page visual documents like slide decks and financial reports.
OpenAI has introduced a new suite of internal security policies designed to contain risk during model development and testing, marking its first major safety overhaul following a July security incident.
The debate over whether artificial intelligence spending has entered bubble territory took center stage as Nvidia CEO Jensen Huang committed $500 billion to independent infrastructure financing platforms while specialized cloud providers report surging revenue.
Engineers led by Hanbat National University have developed a physics-informed AI system that drastically cuts the computational time needed to design phase-change thermal energy storage systems.
Qwen3.8 27B achieved a score of 52 on the Artificial Analysis Intelligence Index v4.1.1, outpacing the median score of 9 recorded for comparable open-weight models.
A participant critique of HackEurope 2026 highlights growing tensions between traditional software engineering and AI-assisted prototyping in startup competitions.
Artificial intelligence startup Anthropic is considering a public market entry that could value the firm at nearly $2 trillion, based on internal forecasts projecting up to $200 billion in annual revenue by 2028.
A technical evaluation of the Qwen 3.8 27B model shows that a 17-gigabyte open-weight model can deliver long-context handling, native vision capabilities, effective tool calling, and competent code generation on consumer hardware.
Anthropic informed prospective investors that its second-quarter revenue surged past $11.5 billion—up more than 14-fold from a year earlier—while achieving positive adjusted operating income ahead of a planned IPO.
Anthropic recorded over $11.5 billion in preliminary second-quarter revenue, marking a 14-fold year-over-year surge as the artificial intelligence lab prepares for a prospective fall IPO.
Anthropic revealed that automated safeguards meant to block bioweapon risks were disabled across vendor platforms for nearly a year, prompting the AI startup to raise its risk assessment level.
Nvidia has proposed cutting its financial backstop for OpenAI's 10-gigawatt Ohio data center project by half, seeking to reduce its financial risk while quarterly filings highlight extensive equity investments in key hardware partners.
A new open-source command-line tool named Mole introduces an automated research agent equipped with strict budget enforcement, verbatim citation checks, and local data privacy guardrails.
A study evaluating cutting-edge AI agents tasked with running independent scientific projects resulted in unambiguous rejections from expert reviewers, highlighting the limits of current automated research capabilities.
Chinese artificial intelligence startup Z.ai disclosed new benchmark results showing its open-weight GLM-5.3 model narrowly surpassed Anthropic's Mythos 5 on the CyberGym vulnerability detection suite.
Alibaba has published the open weights for its Qwen3.8 artificial intelligence models under an Apache 2.0 license, claiming the Qwen3.8-27B variant surpasses Qwen3.7-Plus in performance and software coding.
Klaviyo co-founder and co-CEO Andrew Bialecki detailed how the enterprise marketing platform restructured its operations around AI agents, requiring its entire 2,300-person workforce to achieve Level 3 autonomy.
AI startup Mixedbread has launched Toast 1, a specialized retrieval agent designed to handle context gathering and query decomposition for enterprise language models at significantly lower latency and cost.
The Qwen project has made its 27-billion parameter model available on Hugging Face in FP8 precision, offering machine learning teams lower memory overhead for localized and cloud inference.
Arm co-founder Hermann Hauser warned that while artificial intelligence represents a historic technology revolution, market valuations in some segments have outpaced reality amid risky circular financing arrangements.
Mistral AI has launched a forward-sale program called European Compute Units, securing multi-year enterprise pre-orders to help fund its multi-billion-dollar data center ambitions.
Leading U.S. AI developers OpenAI and Anthropic are lowering prices on intermediate language models as enterprise customers turn to cheaper Chinese alternatives to reduce operational costs.
A new paper published in IEEE Annals of the History of Computing traces how computer science terminology has relied on double meanings since the 1950s, leading the public to confuse statistical classification with human cognition.
New research from Anthropic demonstrates that autonomous AI agents with conflicting goals can escalate into digital turf wars, launching malware against each other and colluding on markets.
OpenAI and AI hardware company Cerebras Systems have introduced Ultrafast Mode, a specialized API tier running GPT-5.6 Sol at up to 750 tokens per second using wafer-scale silicon.
OpenAI Chief Revenue Officer Denise Dresser is stepping down less than a year into her role, marking the second major executive departure at the AI firm this week as it prepares for an initial public offering.
DeepSeek is raising developer API pricing for its V4-Flash and V4-Pro models by up to 1,100% while introducing a peak and off-peak billing schedule starting August 16.
Investors backing Anthropic are eyeing a $2 trillion valuation for an expected autumn public float, setting up what could be the largest public debut of the year despite ongoing regulatory challenges and tightening enterprise AI budgets.
DeepSeek has significantly raised output token prices for its flagship AI models while launching a developer preview of DeepSeek Harness v0.1, entering direct competition with Anthropic's Claude Code.
ClickHouse and Hud have introduced a unified integration linking ClickStack observability with function-level runtime intelligence, creating an automated feedback loop for AI-generated code.
Meta has launched a limited beta test for an on-device AI feature designed to identify fraudulent activity on WhatsApp, as FTC data shows social media scam losses topping $2.1 billion.
Apple is conducting talks with media companies to secure multi-year content deals for Siri AI, proposing a variable payment model based on how often published material is used.
