Cerebras says its CS-4 AI accelerator delivers double the performance on the same chip.
CEO Andrew Feldman called CS-4 “the fastest system in the industry.”
Why it matters: For Cerebras, CS-4 carries two public performance claims: double performance on the same chip and the fastest-system title from CEO Andrew Feldman.
Most published authors have contributed to the development of AI tools that threaten to undermine their livelihoods—without their knowledge or consent.
The source asks whether training AI models on copyrighted books is illegal. It offers no definitive answer, calling the legal question complicated.
Why it matters: Published authors contributed without knowing or consenting to the development of AI tools described as threatening their livelihoods.
OpenAI is calling for California to strengthen SB 53, an AI safety bill the company previously opposed, according to TechCrunch AI.
The company did not support the bill in its earlier position; its new call asks California to make SB 53 stronger.
Why it matters: California lawmakers now face a direct reversal from OpenAI: the company previously opposed SB 53 and is now calling for stronger requirements in the same AI safety bill.
Four teachers told WIRED they became targets of sexualized, AI-generated content, showing that the deepfake epidemic in schools affects educators as well as students.
Their accounts describe the difficulty those teachers faced in finding accountability.
Why it matters: Teachers are targets alongside students: four educators told WIRED that sexualized, AI-generated content reached them while accountability proved difficult to find.
Tesla says Cybercab will launch in 12 days and begin entering robotaxi fleets in Austin and elsewhere. Tesla made the announcement in an August 22 post.
The post says additional announcements will cover rollout schedules and plans, but provides no further details on those plans.
Why it matters: Tesla's August 22 post gives Austin and other locations a 12-day deadline for the first stated step: adding Cybercabs to robotaxi fleets.
MIT Technology Review AI reports that the world now has two known things capable of learning a human language to perfect fluency, up from one four years after ChatGPT's release.
For at least 100,000 years, a human child was the only known example. The supplied excerpt introduces the second example but ends before identifying it or explaining what changed.
Why it matters: Language-acquisition researchers now have a second perfect-fluency case to explain, but the supplied MIT Technology Review AI excerpt identifies neither the learner nor the reason the count changed four years after ChatGPT's release.