Cisco looks to distribute AI security agents throughout the network to monitor for malware. - VentureBeat
Bad guys can use AI too. Adversarial generative malware creation to evade antivirus detection sounds...bad? HackerNoon
NSA offers advice for securing AI systems. - The Reg
Microsoft model can create realistic video of human speaking from a single photo and audio, generates with 170ms latency in near-real-time, for all your deepfake scamming ventures. - Gadgets 360
A few image-to-video examples - Twitter
You can also rent out your GPU when you aren't using it for other people's deepfaking requirements. - 404 Media
Mass-hyper-personalized avatars reach out on behalf of candidates in India. Now you can hook avatars up to ChatGPT and have them talk to the user in real-time. - elblog.pl
James Bond Trailer Gets 2.3M Views Despite Being an AI Fake - The Hollywood Reporter
Survey of AI researchers suggests a 50% chance of AI “outperforming” humans in all tasks by 2047, a projection that has moved forward by 13 years compared to forecasts made a year earlier. - HackerNoon
If I'm reading this right, Azure has 5 GW of data center capacity, wants to grow it 50% per year. - Business Insider
Google moves finance roles offshore to free up cash to invest in AI. They don't exactly mince words, do they? Gemini, what are idiomatic expressions for surprisingly brutal candor? - The Reg
A GPT-4-based ophthalmology model was given 87 cases to diagnose, and approached the skills of specialized eye doctors, scoring 69%, while ophthalmology trainees scored 59%, expert ophthalmologists achieved a median score of 76%, and unspecialized junior doctors scored 43%. - The Independent
Teen Emotions Deciphered by AI and Head-cam. Cue teenage girl eye-roll. Neuroscience News
"There's something like a 10–20 percent chance of AI takeover…The most likely way we die involves…we have deployed a lot of AI everywhere…if for some reason, God forbid, all these AI systems were trying to kill us, they would definitely kill us.” - Ars Technica
OpenAI says startups should make stuff that is still commercially viable if OpenAI gets 100x better, because that is their plan. - THE DECODER
Boston Dynamics switches to all-electric from hydraulic actuators, shows freaky new Atlas robot. - TechCrunch
An AI that can design better robots than itself, and then manufacture them, is probably not that far off. - UNILAD
Brave browser folks gear up to compete with Perplexity. - Search Engine Journal
Warriors' Klay Thompson accused of 'brick-vandalism spree' by Elon Musk's confused AI - SF Gate
Sandia installs Intel's neuromorphic computer, which is able to perform some tasks like stream processing 50x faster and 100x more power-efficiently than a GPU. - The Reg
Andrew Ng talks about agentic workflows in using AI, how to become more of a cyborg and integrate AI tools in an iterative task process, as opposed to assigning it ad-hoc self-contained tasks like some kind of primitive centaur. - YouTube
Follow the latest AI headlines via SkynetAndChill.com on Bluesky
