AI news — March 25, 2026
The 11 top stories PulseAugur surfaced that day, ranked by signal across labs, papers, and developer communities.
-
Anthropic hackathon winners show domain expertise can build AI, but demos aren't products
A recent hackathon hosted by Anthropic showcased the power of domain experts in creating AI applications, with a lawyer winning first place for his AI-powered ADU permit assistant. The event highlighted how individuals with specialized knowledge can now build functional software…
-
GM trains scalable driving AI using simulation and VLA models for rare scenarios
General Motors is developing advanced AI systems to tackle the complex challenges of autonomous driving, particularly focusing on rare and unpredictable "long-tail" scenarios. They are employing a combination of large-scale simulation, reinforcement learning, and foundation mode…
-
Edge AI advances with new silicon, Gen AI, and small models in 2026
In a recent podcast episode, experts discussed the evolving landscape of AI at the edge in 2026. They defined the edge as any computing environment outside of a traditional data center, emphasizing its proximity to real-world data capture and sensors. The conversation highlighte…
-
Guide teaches building config file parsers with Parseff combinators
This guide introduces Parseff, a library for building parsers using parser combinators. It demonstrates how to construct a configuration file parser from scratch, explaining concepts like sequencing, choice, and repetition. The tutorial covers handling comments, blank lines, and…
-
Google DeepMind releases toolkit to measure AI's harmful manipulation tactics
Google DeepMind has released new research and a toolkit to measure AI's potential for harmful manipulation, distinguishing it from beneficial persuasion. The study involved over 10,000 participants across the UK, US, and India, focusing on high-stakes areas like finance and heal…
-
Google DeepMind launches Lyria 3 Pro for longer, more customizable AI music generation
Google DeepMind has launched Lyria 3 Pro, an advanced AI music generation model capable of creating tracks up to three minutes long. This new version offers enhanced control over musical composition, allowing users to specify elements like intros, verses, and choruses. Lyria 3 P…
-
Scott Alexander explores arguments for and against a US-China AI pause agreement
Scott Alexander's Astral Codex Ten blog post explores the complex arguments surrounding a potential pause in AI development, particularly focusing on the idea of a bilateral agreement between the US and China. One perspective argues for a mutually enforceable pause, suggesting t…
-
University professors host AI Cafés to foster community dialogue on AI's societal impact
Academics at Auburn University organized public
-
Gary Marcus discusses war, AI, and Sora's future on BBC, Sirius XM
Gary Marcus will be discussing the intersection of war and artificial intelligence in a live conversation with Bloomberg journalist Katrina Manson. He is also scheduled to appear on Sirius XM with Randi Zuckerberg and on BBC World to talk about the future of AI and the perceived…
-
Spotify launches artist profile protection against AI impersonators
Spotify has introduced a new feature called Artist Profile Protection to combat AI-generated impersonations on its platform. This tool allows artists to review and approve or decline releases before they appear on their profiles, preventing unauthorized tracks from accumulating …
-
Climate movement shifts from managing uncertainty to demanding certainty on emissions
The climate movement has shifted from acknowledging uncertainty in climate change predictions to advocating for certainty, a change influenced by the "Merchants of Doubt" narrative. This reframing, which attributes climate change primarily to the actions of fossil fuel industrie…