Good morning. It's Thursday, September 24, and we're covering a proposed ban on superintelligence, YouTube’s push toward AI creator agents, and ChatGPT’s new voice-driven workflows for mobile.

Plus: a new Real or AI challenge and a guest essay exploring the 50mm problem facing AI headphones.

OpenAI is bringing agentic voice features to mobile, letting users trigger tasks such as drafting documents and summarizing emails through voice. Plus and Pro subscribers can use the Work tab for documents, presentations, sites, and Slack summaries. The rollout follows GPT-Live’s July launch and desktop integration, as Anthropic also streamlines mobile-to-desktop workflows.

Uber is deploying up to 500 sensor-equipped Hyundai Ioniq 5s to capture rare driving scenarios for autonomous-vehicle AI. Each car carries 14 cameras, eight lidars, nine radars and onboard computers, targeting edge cases such as debris and unusual traffic control. Uber says six to 12 months of data could support AV developers, with existing deals involving NVIDIA and Wayve.

AI executives tell the UN Security Council that increasingly powerful systems pose risks that require international coordination. Anthropic CEO Dario Amodei calls AI a potential risk to humanity, while OpenAI’s Sam Altman backs government-led oversight. Yoshua Bengio says the dangers are “real and imminent,” as the U.S. and China pursue approaches to govern AI.

_{^{Guest contributor: Jonathan Chien is an audio technology leader, inventor, and product strategist driving innovation in sound and consumer technology.}}

Twenty-four grams inside one ear cup can shape an entire product. In xMEMS’ engineering comparison, a conventional 50mm driver assembly weighs approximately 42 grams. A speaker plate built around our solid-state speaker weighs just 18 grams. That difference can be used to extend battery life, reduce overall weight, or create a thinner ear cup. → Read the full article here.

Sen. Bernie Sanders and Rep. Greg Casar introduced the Ban Artificial Superintelligence Act on September 23, 2026, proposing a permanent prohibition on developing or deploying artificial superintelligence and an immediate pause on advanced AI development. The bill would establish a cabinet-level Department of Artificial Intelligence to regulate frontier AI.

It would also impose penalties of up to 20 years in prison for individuals and allow a “corporate death penalty” for entities that violate the restrictions. The legislation calls for international agreements, allied coordination and export controls aimed at preventing artificial superintelligence development worldwide. → Read the full article here.

YouTube is expanding its AI-powered creator tools with an agent designed to monitor channels and suggest ways to optimize older content. The agent can identify videos gaining renewed relevance or traction, then recommend changes to titles and thumbnails; it can also generate brand pitches using audience and demographic data from a creator’s channel.

The company is also adding AI-generated thumbnails and titles based on uploaded videos, alongside feedback tools for creators’ existing thumbnails. The changes push YouTube’s creator dashboard beyond analytics toward automated content-based decisions. → Read the full article here.

Anthropic says nearly 950 Claude agents autonomously identified a previously uncharacterized enzyme system in bacteriophages after analyzing a large DNA-sequence database over 21 hours and 210 million tokens. The discovery came from an unusual repeating pattern flagged by one agent for human review, with scientists then conducting additional analysis and laboratory testing.

Anthropic compares the system’s underlying machinery to CRISPR, but says it is still investigating what the enzymes actually do, and there is no evidence yet of practical applications. The result is the first reported discovery from Anthropic’s new wet lab as the company expands AI-assisted research into areas including drug discovery. → Read the full article here.

Ema Raises $77M for AI Agents: Startup targets HR, IT and finance workflows, bringing total funding to $140M and expanding its enterprise push.

Comma.ai Faces Federal Probe: NHTSA is investigating five crashes involving its hands-off driving tech, including two that caused three deaths.

Daily AI Users Still Worry: A Gallup survey finds 68% of Americans who use AI daily are concerned about its implications.

OpenAI Extends Cyber Access to Ukraine: Daybreak will help Ukrainian teams find software vulnerabilities and develop fixes to protect infrastructure.

Basecamp Research Raises $140M: AI drug developer reaches an $800M valuation in a Series C led by Section 32.

Thanks for reading. See you next time!

— Matthew Berman, Nick Wentz & the Forward Future Team