Tesla’s Cybercab Bets on Data It Doesn’t Have Yet
As Tesla prepares to launch its driverless Cybercab, the gap between its 380,000 unsupervised miles and rivals’ hundreds of millions raises hard safety questions.
8 Sep 2026
Gemini 3.1 Pro Still Carries Google’s Pro AI Flag as 3.5 Delay Drags On
Google DeepMind’s Gemini 3.1 Pro remains the default Pro-tier model for enterprises in mid-2026, even though Gemini 3.5 Pro was announced in May and has yet to ship.
8 Sep 2026
Waymo Robotaxi Loops for Five Minutes With Passenger Unable to Exit
A software glitch trapped a Waymo passenger in a five-minute loop near LAX, raising urgent questions about remote intervention, passenger control, and regulatory oversight as robotaxi fleets expand.
8 Sep 2026
Inaccessible 3D-Printable Delivery Robot Listing Shows Platform Verification Gaps
A Cults3D model for a toy autonomous delivery rover returns a 403 error, leaving only unverified secondary descriptions. The case highlights risks for professionals sourcing open 3D-printable designs.
8 Sep 2026
Huawei Unlikely to Catch Nvidia by 2030, Epoch AI Finds
New roadmap analysis shows Huawei's Ascend 950 delivers half the performance of Nvidia's H100 and total compute output under 4% of Nvidia's in 2026. Rising Chinese demand for Ascend chips does not signal technological parity.
8 Sep 2026
OpenAI launches GPT-6 Astra amid breakthrough math and safety alarms
The model solved ten previously unsolved problems, including three Erdős problems, but its opaque reasoning and critical cyber risk classification raise urgent governance questions.
8 Sep 2026
Google’s Gemini grows from chatbot to robotics, CLI and speech benchmarks
Google’s multimodal model family now spans tiered releases, robotics vision-language models, developer CLI tools and spoken-language research. The expansion positions Gemini as an embedded platform rather than a single chatbot.
7 Sep 2026
LLM Daily: September 06, 2026
OpenAI’s GPT-6 “Astra” model, launched on September 5, 2026, was successfully jailbroken within 24 hours, according to the September 6, 2026 edition of LLM Daily, a newsletter published on Buttondown.
7 Sep 2026
Google I/O 2026: Gemini becomes an always-on agentic platform
Neural Expressive redesign, Gemini Spark 24/7 agent, Gemini 3.5 models, and a $100 AI Ultra plan push the assistant from chatbot to proactive workflow engine.
7 Sep 2026
When 99% on a Benchmark Means Nothing
The old tests that crowned AI models are saturated. New evaluations focus on real tasks, but no single number predicts production success.
6 Sep 2026
LLM Benchmarks in 2026: The 7 Evaluations That Matter
As MMLU saturates, model selection now depends on harder, contamination-resistant evals—from GPQA to SWE-bench Pro. Here are the seven benchmarks and evaluation strategies that define frontier AI in 2026.
6 Sep 2026
AI Video Generation in 2026: No Single Model Wins Every Task
ByteDance’s Seedance 2.0 leads text-to-video, Google’s Gemini Omni Flash tops image-to-video, and Alibaba’s Happy Horse 1.0 wins video editing, showing task-specific strengths and cost implications.
6 Sep 2026Newsletter
Get the AI news that matters
One short brief with the day's most important AI stories — written for professionals.
We send a confirmation link. No spam. Unsubscribe anytime.