Between two big titles, several news caught our attention this week without being the subject of an article. Here is an overview.
Meta trial: rejected complaint, but debate on the revived “fair use”
A federal judge rejected the complaint of authors, believing that they had not demonstrated that Meta’s AI compromised the economic outlets of their works. However, he admitted that the use of protected works to cause AI could be illegal in many cases, despite the doctrine of the fact invoked by technological companies.
Cloudflare blocks the most massive ddos attack ever thwarted
Cloudflare blocked the largest DDOS attack ever recorded in debit: 7.3 TB/s, or 37.4 Teraoclets sent in 45 seconds, the equivalent of 9,350 HD films or 9.35 million songs downloaded in less than a minute.
Lakera, EPFZ spin-off, launches AI Model Risk Index
Lakera, EPFZ spin-off, launched AI Model Risk Index, a new standard to assess LLM safety. This system measures the level of real risk under attack conditions, simulating concrete scenarios to guarantee the reliability of AI in various sectors, from health to health.
Gemini Robotics On-Device performs AI locally on robots
Google unveiled Gemini Robotics On-Device, an optimized version of its VLA (Vision-Language-Action) model for bi-bras robots. The latter would offer great dexterity, rapid adaptation to tasks and network without network. An SDK facilitates its integration and experimentation.
