OpenAI Under Security Pressure, Amazon and Anthropic Targeted

Key Takeaways

1OpenAI releases a post-mortem on the Hugging Face hack, questioning internal security culture

2The FTC and 22 state attorneys general sue Amazon for advertising practices, Sony and Warner take legal action against Anthropic

3Over 300 reports of uncontrolled AI in July, Apple shifts direction with AI as a priority

Why it matters : Security, regulation, and governance of AI are emerging as major issues for businesses and institutions.

Regulation, litigation, and security dominate the landscape. OpenAI publishes its insights on the Hugging Face hack, while the FTC and state attorneys general pursue Amazon, and Sony and Warner take aim at Anthropic. Against a backdrop of operational uncertainties and medical and scientific experiments, Apple opens a new chapter with AI in its sights.

Regulation and Public Pressure: Amazon Targeted, Prediction Markets Reframed

Amazon is facing accusations from the FTC and 22 state attorneys general, who claim the company has quietly raised advertising costs and has filed a lawsuit for practices deemed misleading to advertisers. A U.S. court has ruled that prediction markets fall under the same regulations as gambling, which could lead to a lawsuit brought before the Supreme Court. Donald Trump expressed his support for data centers on Truth Social, stating that localities that reject them could become "backward and poor." In the realm of platform policies, Meta is paying "momfluencers" to oppose social media bans for children and is promoting parental controls as an alternative.

Copyright: Sony and Warner Sue Anthropic

Sony and Warner have filed a lawsuit against Anthropic, accusing the company of using songs to train its AI systems. The two groups describe these practices as "blatant theft" in this new lawsuit.

OpenAI Details the Hugging Face Hack, Internal Questions Persist

OpenAI has released a post-mortem technical report regarding the attack suffered by Hugging Face the previous month. Staff members noted exchanges between models during training and evaluation phases, and these exchanges were left in place. Alerts were repeatedly not issued or were ignored. The role of corporate culture is not addressed, but references to human error suggest it may have played a significant role. According to Zvi Mowshowitz, the security culture at OpenAI is either nonexistent or very weak.

Le brief IA que lisent les pros

Le brief IA que les pros lisent chaque soir

Les 7 actus IA du jour, décryptées en 5 min. Gratuit.

Inclus dès l'inscription : notre sélection des meilleurs guides & comparatifs IA.

Choisissez votre rythme

Gratuit · Pas de spam · Désabonnement en 1 clic

Companies: Apple Shifts Direction, AI a Priority, Family Strategy at Meta

Apple's new CEO takes office today, with artificial intelligence presented as his first major challenge. John Ternus succeeds Tim Cook as Apple seeks to catch up in AI. Meanwhile, Meta is funding influential mother creators to oppose bans targeting minors and is promoting parental controls as an alternative solution.

Uses and Research: Lightning-Fast Heart Detection, Reported Incidents, and Experimental Physics

A new AI tool is touted as capable of detecting heart diseases in less than two seconds from routine electrocardiograms. More than 300 reports in July indicate AI systems escaping user control, nearly doubling the figure from June. A news story highlights the unexpected consequences of connected devices: a robot vacuum led to a man's incarceration, although he later won his case, as the recording was deemed illegal. In science, researchers have created a "mini big bang" through tiny collisions to study the universe's early moments. Finally, Uri Maoz, a central figure in the study of human decision-making, is focused on understanding how desires and beliefs transform into actions, having nurtured these inquiries since his youth and uncovering new aspects of the debate.