01
Bloomberg Technology ↗
Anthropic and OpenAI's warnings about adversarial distillation, the practice of querying a rival's AI model at scale to train a cheaper competitor on its outputs, are reviving a long running Silicon Valley debate now playing out in Washington. Anthropic says China operated roughly 25,000 fraudulent Claude accounts over six weeks that harvested 28.8 million exchanges with its models, and has previously accused several Chinese AI labs of running similar schemes to mine its outputs. The company is now pressing lawmakers to treat the practice as a national security issue rather than a private contract dispute, calling for penalties and stronger export controls, with OpenAI, Anthropic, and Google reportedly beginning to coordinate on curbing it.
02
Bloomberg Technology ↗
Joe Lonsdale, managing partner at 8VC and co-founder of Palantir, says AI is fueling a new era of venture investing, discussing his firm's record $1.5 billion fund and why startup funding rounds are getting bigger. He told Bloomberg Tech he is placing bets on the next generation of defense tech and AI companies as capital continues pouring into the sector.
03
Next Big Future AI ↗
XAI Grok 4.5 is benching as good as or better than Anthropic Opus 4.7 and 4.8 at about half the cost. After only five days it is tracking to about 15 percent of Anthropic Opus 4.8 usage per Openrouter, and usage is still growing.
04
Bloomberg Technology ↗
Intel is spending 5 billion euros, about 5.7 billion dollars, to expand its Leixlip campus in Ireland, upgrading existing fabrication lines rather than building new plants. The investment will add capacity for Intel Xeon 6 and next generation Xeon chips built on the Intel 3 process node to meet surging AI and high performance computing demand. Intel says the expansion will create several hundred permanent jobs plus thousands of construction jobs and should finish by the end of next year, bringing the company's total investment in Ireland since 1989 to more than 30 billion euros.
05
TechCrunch AI ↗
Anthropic has begun rolling out rupee-denominated subscription pricing for Claude users in India, which has become the company's largest market outside the United States. The move mirrors localized pricing strategies already used by rivals such as OpenAI in high-growth markets, and signals Anthropic's push to compete more aggressively on cost in Asia.
06
Forbes Tech ↗
A Science Advances study of 15 million biomedical abstracts found signs of large language model involvement in about 13.5 percent of 2024 papers, prompting publishers including Elsevier and Springer Nature to bar AI from being listed as an author and funding bodies like NIH to restrict reviewers from using generative AI on proposals. In response, an open source Academic Humanizer tool has emerged that strips telltale AI phrasing, such as generic transitions like paves the way, and rewrites text to match an author's established voice. The core tension publishers face is separating legitimate AI assisted editing, which helps non native English speakers, from AI generated fabricated citations or unsupported claims that undermine research integrity.
07
Next Big Future AI ↗
New testing suggests the high reasoning tier is the sweet spot for both OpenAI GPT 5.6 Sol and Anthropic Claude Fable 5. GPT 5.6 Sol excels at agentic coding and long horizon tasks but is far more persistent than its predecessor, sometimes burning over 15 percent of a 5 hour usage window through aggressive sub agent spawning, so OpenAI recommends sticking to High rather than Maximum reasoning. Claude Fable 5, by contrast, is strong at reliable parallel sub agent orchestration, making it better suited to act as a commander that delegates mechanical work to cheaper sub agents. For both models, higher reasoning tiers beyond High show diminishing returns relative to their added cost.
08
CNBC Technology ↗
Meta's Hyperion data center supercluster in Richland Parish, Louisiana, will be a 5 gigawatt facility now expected to cost more than 50 billion dollars, up sharply from the 27 billion dollar figure disclosed last October when Meta and Blue Owl Capital formed a joint venture to fund and manage the build out. A 2024 Louisiana law exempts data centers built before 2029 from state and local sales tax on equipment for 20 years, and local officials say the resulting tax revenue is funding community improvements including higher teacher bonuses. Meta has also pledged 1 billion dollars for local infrastructure like roads and water systems, with the full 5 gigawatt buildout targeted for completion around 2032.
09
TechMeme ↗
In a Verge Decoder interview, Nvidia automotive chief Xinzhou Wu said Level 4 autonomy could reach consumer cars in under five years, backed by the company's roadmap of a 2025 Mercedes Benz production project, a 2027 Level 4 pilot with Google, and driverless service plans with Uber for the 2028 Los Angeles Olympics. Wu argued that cameras, radar, ultrasonic sensors, and lidar together provide the redundancy needed to handle safety critical edge cases, calling lidar indispensable for Level 3 and Level 4 systems. Asked about Tesla's camera only approach, Wu acknowledged Tesla is ahead on basic driver assistance but stopped short of saying vision alone can reach full Level 4 autonomy.
10
Business Insider Tech ↗
Microsoft first invested 1 billion dollars in OpenAI in 2019 and deepened the relationship in 2023, with CEO Satya Nadella backing Sam Altman when he was briefly ousted by the board, but by this April the two companies had effectively broken up and converted their exclusive arrangement into a non exclusive one following reports Microsoft considered suing OpenAI for breach of contract. Apple is now suing OpenAI, alleging Altman's company poached its employees and stole trade secrets, and OpenAI executives were reportedly disappointed enough in the Apple partnership to consider a countersuit of their own. The pattern points to a recurring dynamic where OpenAI's biggest technology partnerships curdle into legal disputes as the company grows.