Politico: Sources: Meta decided to support KOSA after it was packaged with language preempting state AI laws and a bill requiring app stores to verify users' ages  —  A children's online safety bill at the center of the White House's push to preempt state regulations on artificial intelligence has an unlikely new ally: Meta.

Discussion | Link

Discussion | Link

Discussion | Link

Matt G. Southern / Search Engine Journal: Kapwing: in tests, ~60% of the first 500 TikTok For You videos shown to a new account were AI slop, ~3x the slop shown to a new YouTube user via YouTube Shorts  —  In a fresh-account test, Kapwing found 59% of TikTok For You videos were AI slop, roughly three times the rate on YouTube.

Discussion | Link

Mukesh Sharma/Unsplash If someone used a large truck to deliver one envelope across the street, what would your reaction be? You would probably say it worked, but it was wasteful. The envelope arrived, but the method made little sense. In many ways, this mirrors how we often use artificial intelligence (AI) today. We use powerful AI systems such as ChatGPT to write short messages, polish simple sentences, and answer questions we could handle ourselves with less intensive tools. The outcome is the same. But under the hood, the differences in processing, electricity, and water use are enormous. One small request will not change the planet. But millions of small and unnecessary requests can add up over time. The question is not whether AI is good or bad. The better question is whether we are using the right amount of AI for the right task. AI is becoming a utility AI is quickly becoming an everyday utility, like electricity. But because it feels invisible, we can easily forget every use has a cost. We know not to leave lights on all day. We also know not to run an air conditioner with windows open. We need a similar mindset for AI. AI is often described as living in “the cloud”. This makes it sound light, clean, and almost magical. But the cloud is not really a cloud. It is someone else’s computer. More precisely, it is a large network of data centres filled with servers, chips, cables and cooling systems. When we interact with AI systems, they do not simply “know” the answers. They run calculations through large computer systems to generate a response. Training large AI models require enormous resources, but daily use adds up too. Data centres can create significant carbon emissions. They also take up large tracts of land, put pressure on electricity grids and water supplies, and generate electronic waste. The International Energy Agency has projected global electricity use by data centres could roughly double to around 945 terawatt-hours (TWh) by 2030. How to use AI more sensibly This makes it important to consider how we can use AI more sensibly – and sustainably. First, choose the right tool for the task. Not every job needs the most powerful AI model. Sometimes a smaller tool, search engine, calculator, or even your own brain is enough. Ask yourself: is AI adding any value here, or am I using it out of habit? Second, write clearer prompts. Vague questions often lead to many follow-up prompts. A clear first request can reduce this back and forth. It is like giving a taxi driver the correct address before the trip starts instead of constantly correcting the route along the way. Third, ask only for what you need. If you need a short list, do not ask for a long report. More words usually mean more computing. Fourth, be more careful with images, audio, video and other media files. These typically require more computing power than texts. The role of organisations and governments Large organisations have an even bigger responsibility. They should not add AI to every product just because it is fashionable. Before adopting AI, they should ask: what problem are we solving? Is AI really needed? If it is, can we use a smaller model? Governments should also require large data centres to report electricity use, water use, emissions and electronic waste as part of planning approvals. Australia has released guidelines for data centre developers, but they are voluntary, not law. New data centres should be planned very carefully, especially in places with limited water or stressed grids. Sustainability labels, like energy ratings on appliances, could help users compare tools by efficiency. These measures could make sustainability part of normal AI governance, not an afterthought. They could help ensure AI stops being the equivalent of the truck carrying one envelope across the street. Seyedali Mirjalili does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.

Karen Freifeld / Reuters: Sources: the US has held off adding DeepSeek, CXMT, and 100+ other flagged companies to its Entity List, last updated in October, the longest gap in 10+ years  —  The U.S. has held off adding China's AI startup DeepSeek, memory chipmaker CXMT and more than 100 other companies flagged …

Marketing technology has seen plenty of change over the past few decades. But what...

Bloomberg: Sources: Amazon faces a possible FTC lawsuit alleging it misled advertisers; multiple state AGs are involved and the case could seek billions in civil penalties  —  Amazon.com Inc. is facing a possible lawsuit from the US Federal Trade Commission that may lead to billions of dollars in civil penalties …

Politico: Many US lawmakers said this week they had not been briefed on Trump's Fable 5 move, including Ted Cruz, whose commerce committee has jurisdiction over AI policy  —  Members of both parties said they now see an opening to mobilize their colleagues around legislation that would reclaim congressional authority …

Mitchell Armentrout / Chicago Sun Times: Illinois Governor JB Pritzker signs a state budget that levies taxes on social media giants, digital advertising, and prediction and crypto markets  —  The eighth state spending plan of Pritzker's tenure includes additional funding for food assistance in the wake of deep federal funding cuts under President Donald Trump.

Introducing LifeSciBench, an expert-authored, expert-reviewed benchmark for evaluating how AI systems handle real-world life science research tasks and decisions.

The Senior Fellow, Global Programs, will lead a tightly scoped, policy-responsive workstream at the intersection of AI, industrial policy, and global political economy; building on AI Now’s existing research on AI nationalism and translating it into a directed research and policy agenda. The terrain around “AI sovereignty” is rapidly being reshaped by an aggressive US […] The post AI Now is Hiring a Senior Fellow, Global Programs appeared first on AI Now Institute.

Snap unveiled its Specs AR glasses with EyeConnect at a Long Beach conference. Joe Scarnici/Getty Images for Snap Snap unveiled its Specs AR glasses with EyeConnect at a Long Beach conference. The internet is balking over the $2,195 price tag on the Specs. The AR business has been struggling with profitability, and Meta has already lost billions. After more than a decade of work, Snap on Tuesday unveiled Specs, a new pair of augmented reality glasses with a price tag more commonly seen on a gaming PC or high-end TV: $2,195. The launch marks Snap's biggest bet yet on a future where computing moves from smartphones to eyewear, a sentiment frequently shared by Mark Zuckerberg, who made the prediction that smart glasses will become the next major computing platform, much as smartphones displaced PCs as the primary way people access digital services. Still, Specs cost far more than Meta's Ray-Ban smart glasses, which start around $350, though they're cheaper than Apple's $3,500 Vision Pro headset. The usual AR features are expected: wearers will be able to record first-person footage, browse the web, and play games, along with a feature called "EyeConnect," which lets two wearers launch shared experiences simply by making eye contact. Still, that might not be enough of a differentiator given the massive price tag. "Snap Specs for $2,195 is an instant nope for 99% people," said Ray Wong, senior editor of Gizmodo, on X. Raj Nijjer, the CMO of Symmetry Software, called the Specs "out of touch," while Riley Brown, a tech educator and startup founder, posted on X that the Specs would be "dead on arrival." Each pair of Specs also weighs around 132 grams, nearly twice that of Meta's Ray-Bans, which weigh around 70 grams. Snap's Specs cost $2,195 and will ship by fall 2026. Phillip Faraone/Getty Images for Snap Specs arrive as Meta has established an early lead with its Ray-Ban partnership, while Google recently unveiled its own AI-powered eyewear plans. A problem dogging the tech is that no one has cracked the economics. Meta's Reality Labs division, the unit responsible for its VR headsets, AR glasses, and metaverse ambitions, has become one of the most expensive bets in tech. Since 2020, the division has racked up tens of billions of dollars in operating losses, including more than $17 billion in 2024 alone. Snap has also been facing mounting challenges. The company has spent years trying to turn its AR vision into a sustainable business while navigating a volatile advertising market, slowing user engagement in the US, and a stock price that remains well below its pandemic-era highs. Earlier this year, Snap spun off a separate company focused on bringing Specs to market. In April, Snap also conducted a round of layoffs as it continued efforts to streamline operations and control costs. If you're convinced Specs are worth the price tag, preorders open on June 16 with a $200 refundable deposit, and shipping will begin this fall in the US, UK, and France. Read the original article on Business Insider

Anthropic's popularity with business users is growing so well that the latest beef with the government might actually boost it, data from Ramp suggests.

Developers building for AR glasses and wearable devices face an infrastructure gap. The hardware is ready, but creating AI experiences requires integrating live... Developers building for AR glasses and wearable devices face an infrastructure gap. The hardware is ready, but creating AI experiences requires integrating live camera and microphone streams, multimodal AI models, enterprise data, tool use, deployment infrastructure, and device-specific runtimes. NVIDIA XR AI is designed to address this challenge by providing a reusable foundation for… Source

In addition to executing entirely in memory, the malware's infection chain incorporates other anti-analysis techniques designed to evade detection.

NAACP lawsuit says xAI uses gas turbines without permits for Grok data center.