Google Bows New Cheaper AI Super Chips; UK-100 Countries Can Hack Your Phone; OpenAI’s New Image Generator Can Search Web-Multiple Images; Unauthorized Users Already Into Anthropic’s Mythos AI

Google has rolled out new AI ’super chips’ which then say are cheaper and at the same time give faster responses than rivals like Nvidia. Benzinga.com reports that Google VP Mark Lohmeyer says “The number of transactions is going way up, and the cost per transaction needs to go way down for it to scale.” The company will offer TPU 8t for building AI models and TPU 8i for running them, aiming to improve performance and cost efficiency. Google continues to partner with Nvidia, despite being a rival for chips. The two companies have teamed up for over a decade now. Google says it will continue to work to make it easier and cheaper for companies to build and run AI applications. 

The United Kingdom’s National Cyber Security Centre says more than half the world’s governments have access to commercial spyware that can break into computers and phones to steal sensitive information. According to techcrunch.com, a report being released by the UK government suggests that the barrier to access this type of surveillance technology has fallen, potentially making it easier for foreign governments and hackers to target U.K. citizens, companies, and critical infrastructure with spyware. Of course, if they can get into the devices of UK citizens or companies, they can just as easily do so with American citizens and to US based companies. Over 100 countries now have access to commercial spyware developed by the likes on NSO Group…that makes Pegasus, and also Paragon’s Graphite. So far, most of the activity…at least according to the UK government…is coming from adversarial governments and not cybercriminal gangs. The net has widened, however…more bankers and wealthy businesspeople are being hacked. I guess not being wealthy makes me relatively safe…for now.

OpenAI is in the midst of rolling out its latest AI powered image generator, which they claim has ’thinking capabilities,’ giving it the ability to search the web to help it create multiple images from a single prompt. Theverge.com says that according to OpenAI, ChatGPT Images 2.0 can create more ’sophisticated’ images, and can pull information from the web. It can also created visual explainers based on files you upload. The model can create up to 8 images at once with thinking enabled. OpenAI says this should make it easier to generate things like manga pages, a series of social graphics, or design plans for every room in a house.

Who didn’t see this coming…I hope including people at Anthropic. A small group using a private Discord got into the Claude Mythos Preview model the day it was announced. Thenextweb.com notes that Anthropic says it is investigating and has found no evidence of impact to its core systems. This model is the tightly restricted cybersecurity model Anthropic briefed the White House on, and which the NSA is already using. The group managed to guess the url and got into Mythos. The group, whose members communicate via a private Discord channel dedicated to gathering intelligence on unreleased AI models, has been using Mythos regularly since gaining access and provided Bloomberg with proof in the form of screenshots and a live demonstration. Anthropic has assured that the group hasn’t penetrated cores systems. A person currently employed at a third-party contractor working with Anthropic appears to have been involved, at least in part, in facilitating the group’s access.

I’m Clark Reid and you’re ‘Technified’ for now.