OpenAI is revising its Pentagon contract after Sam Altman called the deal 'opportunistic and sloppy' and said the company is working with the Department of Defense on provisions addressing mass surveillance. - Financial Times

OpenAI amended its Pentagon contract to prohibit intentional domestic surveillance or tracking of U.S. persons, bar monitoring using commercially acquired personal data, and received Pentagon assurance the technology will not be used by defense intelligence agencies, including the NSA. - The New York Times

OpenAI CEO Sam Altman said the company "shouldn't have rushed" its DoD deal and announced contract language barring intentional domestic surveillance of U.S. persons, a Department affirmation agencies like the NSA won't use OpenAI tools, and technical safeguards. - CNBC

What’s OpenAI going to do when the truth comes out? - Platformer

How OpenAI caved to the Pentagon on AI surveillance. The law doesn’t say what Sam Altman claims it does. - The Verge

Sam Altman statement - X (formerly Twitter)

Sam Altman posted a statement saying OpenAI should prioritize alignment, democratization, empowerment and privacy, avoid deciding the fate of the world alone, collaborate with governments without drastic regulatory overreach, and pursue iterative deployment and public education to manage AI risks. - X (formerly Twitter)

ChatGPT mobile uninstalls spiked 295% within 48 hours after OpenAI confirmed a partnership with the U.S. Department of Defense, prompting account deletions and a surge in rival Anthropic Claude downloads. - techputs

An alleged leak of OpenAI's GPT-5.4 emerged during a Codex demo, while GPT-5.3 is already integrated into Codex, and OpenAI has not confirmed the leak, raising concerns about security and protection of proprietary models. - Geeky Gadgets

The Pentagon ended its partnership with Anthropic after a disagreement described as a 'fight about vibes'. - The Wall Street Journal

The U.S. Treasury, State Department and Federal Housing Finance Agency are terminating use of Anthropic products, with the State Department switching its StateChat to OpenAI's GPT-4.1 after President Trump ordered a six-month phase-out of Anthropic. - Reuters

Anthropic Pitched AI for Drone Swarms Amid Pentagon Contract Dispute - Bloomberg

The Pentagon moved to cut ties with Anthropic, labeling it a 'supply chain risk' when talks over banning autonomous weapons collapsed; OpenAI reached a similar Pentagon deal and U.S. Central Command reportedly used Anthropic's Claude, Emelia Probasco said. - New York Magazine

Anthropic, whose research scientist Amanda Askell helped author a "Claude constitution," has resisted Pentagon requests and released ethics guidelines, while critics warn against anthropomorphizing Claude and urge Congress to regulate and limit military or societal-scale ethical uses of LLMs. - The Dispatch

NY Post does a deep dive into Amanda Askell's philosophical blog to tag Anthropic as leftwing nut jobs. - New York Post

basically the war-on-truth folks have Anthropic in their sights, while using it to bomb everything in sight

Anthropic added memory to Claude's free tier, after launching it for paid users in October 2025, enabling the chatbot to reference past conversations and allowing users to pause or permanently delete saved memories. - Engadget

Anthropic's Claude was reportedly used twice by the Trump administration—first in planning an attempted abduction of Nicolás Maduro and later to parse intelligence to aid targeting in a strike on Iran—prompting industry conflict and calls for international AI weapons limits. - the Guardian

US forces used Anthropic's Claude in a Palantir–Pentagon system to accelerate strike planning, as almost 900 US and Israeli strikes in the first 12 hours included a missile that hit a school, killing 165 people, the UN said. - the Guardian

U.S. military units have begun using commercial chatbots for combat operations, reportedly employing Anthropic’s Claude in an operation to capture Venezuelan president Nicolás Maduro and in a later U.S. strike that killed Iran’s leader Ayatollah Ali Khamenei. - Fast Company

Anthropic's AI Claude surged to No.1 on Apple's US free app chart and climbed Android/UK charts after the Pentagon effectively blacklisted Anthropic over ethics concerns, with free users up >60%, daily sign-ups quadrupling and paid subscribers more than doubling. - the Guardian

Death is a process, and the Republic is dead. The government basically asserts the power to destroy Anthropic over practically nothing, except standing up for a principle. Take a moment to mourn the republic that was. - Hyperdimensional

US politicos are all as dumb as Trump, they get their info from 3rd grade picture books - Axios

Meta is testing an AI shopping research tool intended to compete with OpenAI's ChatGPT and Google's Gemini. - Bloomberg

Google updated Google Home with Live Search, letting Gemini analyze live camera feeds, added improved context-aware room targeting and media playback, and offers Gemini features via Google Home Premium Advanced subscription ($20/month or $200/year). - The Verge

Tech giants take on record debt to fund AI data center expansion - Financial Times

Nvidia is investing $2 billion each in Lumentum and Coherent, $4 billion total, in multiyear nonexclusive deals with purchase commitments and capacity rights to secure photonics components for optical interconnects addressing AI data-center bandwidth, latency and power constraints. - The Verge

A study finds AI data centers may not need continuous peak power, indicating operators could provision lower average capacity rather than constant peak capacity. - Bloomberg

Huawei is offering prefabricated 'flatpack' AI datacenters based on its Kunpeng CPUs and Ascend GPUs, claiming site prep in 4–6 months and 1,024-node super-clusters deployable within 15 days, while tests show its silicon trails Intel/AMD CPUs and Nvidia GPUs. - The Register

U.S. prosecutors in December 2025 alleged a smuggling ring exported or attempted to export about 7,000 Nvidia H100/H200 GPUs (≈$160 million) to China, charging defendants including Fanyue Gong, Benlin Yuan and Alan Hao Hsu. - The Wire China

Researchers reverse-engineer Apple's M4 Neural Engine, revealing 16-core graph accelerator with 2-4x throughput gains bypassing CoreML - The Potential Surface

Qualcomm launched a Dragonwing robotics processor, and CEO Cristiano Amon said robotics will become a 'larger opportunity' within two years, as analysts forecast a $370 billion robot market by 2040 and $9 trillion humanoid market by 2050. - CNBC

Cursor run rate doubles from 3 months earlier. - Bloomberg

Nvidia-backed, NYC-based Reflection AI seeks $20B valuation to rival DeepSeek - Financial Times

OpenAI announced multi‑year partnerships with BCG, McKinsey, Accenture and Capgemini to deploy AI, while Anthropic sealed deals with Accenture and Deloitte, which launched a 15,000‑staff Claude certification; McKinsey says about 40% of its work is AI-related. - Business Insider

BNY executive reports measurable operational gains from AI implementation - Bloomberg

Deepfake attacks surged in 2025—over 8 million deepfakes and $1.1 billion stolen from U.S. corporate accounts—while only 32% of executives feel prepared, prompting calls for coordinated disclosure and crisis protocols across legal, cyber and communications. - AOL

Organizations are combating AI-powered phishing—which uses large language models, automation and voice synthesis to craft personalized attacks—by deploying AI email security, Zero Trust, phishing-resistant MFA, continuous training and UEBA, though costs and talent constraints limit smaller firms. - C# Corner

Ars Technica retracted a Feb. 13 story after discovering AI-fabricated quotations attributed to engineer Scott Shambaugh; co-author Benj Edwards said he used Claude Code and ChatGPT while ill, inadvertently inserted misquotes, and was fired after an internal review. - Futurism

not a great look but, get your news from places that fire people if they make things up, rather than places that fire people if they don't. (like most people these days it seems)

An Inside Higher Ed/Generation Lab poll found about 85% of undergraduates use generative AI for coursework and about 19% use it to write full essays, while professors debate and adopt divergent policies and classroom practices. - NPR

The Supreme Court declined to review Stephen Thaler's appeal, leaving intact lower-court and appeals-court rulings and a Copyright Office determination that the 2019 AI-generated image 'A Recent Entrance to Paradise' lacks human authorship and cannot be copyrighted. - The Verge

AI will not replace strategic decision-making but will expose weaknesses and assumptions in existing strategies. - Fast Company

Anthropologists including Michael J. O'Brien trace two million years of technological traditions and argue technological progress scales superlinearly through specialist collaboration, recommending AI be tuned to find and trust expertise rather than dilute it to accelerate innovation. - Fast Company

you are not serious people - Instagram

thank you for your attention to this matter. - Instagram

Follow the latest AI headlines via druce.ai on Bluesky