🌐 DeepSeek V4 falls short of top US AI models in benchmarks

😔 OpenAI CEO apologizes for not flagging a shooter to police

🤖 Anthropic tested AI agents buying and selling real goods

🔋 EU mandates replaceable phone batteries by 2027

📱 Samsung warns it could post its first-ever smartphone loss

+ 📊 Daily poll and results

+ 📈 Trending tools and resources

1,000+ ready-to-use prompts to solve problems in minutes instead of hours—tested & used by 1M+ professionals

Superhuman AI newsletter (3 min daily) so you keep learning new AI tools & tutorials to stay ahead in your career—the prompts are just the beginning

DeepSeek's V4 Pro scored 52 on the Artificial Analysis Intelligence Index, behind OpenAI's GPT-5.5 at 60 and both Claude Opus and Gemini 3.1 Pro at 57, leaving the US-China AI gap roughly where analysts estimated it: about seven months.

The model does show real progress, combining a one-million-token context window with compatibility for Huawei's Ascend chips and a 90% reduction in KV cache usage, but experts flagged that DeepSeek never disclosed whether restricted Nvidia Blackwell chips were used in training.

Markets noticed the underwhelm: Nvidia stock rose 4.32% on the V4 news, a sharp contrast to the 17% single-session crash it suffered when DeepSeek's R1 dropped.

OpenAI banned 18-year-old Jesse Van Rootselaar's ChatGPT account in June 2025 after she described gun violence scenarios, but staff debated and decided against contacting police before she allegedly killed eight people in Tumbler Ridge.

Sam Altman published an apology in the local Tumbler Ridge newspaper, saying he discussed the shooting with the mayor and BC Premier, who all agreed a public apology was necessary after a period of grief.

British Columbia Premier David Eby called the apology "necessary, and yet grossly insufficient" — and Canadian officials are now weighing new AI regulations, though no decisions have been made yet.

Anthropic ran "Project Deal," an internal marketplace where AI agents negotiated on behalf of 69 employees given $100 budgets, producing 186 deals worth over $4,000 in total value.

Testing four marketplaces with different models, Anthropic found users represented by more advanced AI got measurably better outcomes, while users on the losing end had no idea they were outbargained.

Anthropic flagged this as a genuine concern: if agent quality gaps spread to real commerce, people could routinely lose negotiations without ever knowing a better model beat them.

Starting February 2027, phones sold in the EU must have batteries users can swap without specialized tools, unless manufacturers provide those tools free with the device.

The law exempts phones with batteries that hold 80% capacity after 1,000 charge cycles. iPhone 15 and newer already qualify, as do Google Pixel 10 Pro models, so most flagship phones may skip redesigns entirely.

Samsung and Nothing actually exceed the exemption bar, rated for 1,200 and 1,400 cycles respectively, which makes the law's practical impact on the industry narrower than the headline suggests.

Samsung's mobile chief has warned company leadership that the AI-driven surge in DRAM and NAND prices could push the division to its first net loss ever, even as Galaxy S26 sales stay strong.

AI infrastructure is the culprit: a single Nvidia Vera CPU server consumes as much LPDDR5x RAM as 4,600 Galaxy S26 Ultras, pulling memory supply away from consumer devices and roughly doubling memory costs in smartphone bills of materials.

Samsung's semiconductor division just posted an estimated $38 billion profit in Q1 2026, so the company isn't hurting overall — but that's cold comfort for anyone eyeing the $50-to-$100 price hikes already hitting Samsung's mid-range and foldable lineup.

A) Musk — the nonprofit betrayal case is real - 50%

B) Altman — OpenAI's evolution was never illegal - 50%