Some of the staff at Character.AI have moved to Google, chief among them the young company’s co-founders, Noam Shazeer and Daniel De Freitas.
Founded in 2022 and headquartered in Menlo Park, California, Character Technologies Inc. specializes in what it calls personalized AI and is working towards artificial general intelligence (AGI) – in short, AI that outperforms human beings at many tasks. Its direct-to-consumer product is personas for the AI to speak through with users. These personas, or characters, can be selected or created, and they can be inventions or imitations of real people – moguls like Elon Musk, for instance, or athletes like Lebron James. There are multiple versions of Albert Einstein.
Shazeer and De Freitas have both joined Google Deepmind – the one in LLM development, the other as a research scientist, according to their respective LinkedIn pages. Back at Character.AI, meanwhile, General Counsel Dominic Perella has stepped in as Interim CEO, to replace Shazeer. Perella was for seven years Vice President, Deputy General Counsel and Chief Compliance Officer at the software developer Snap.
In its early days, says Character.AI, “personalized superintelligence required a full stack approach,” with the pre- and post-training of large language models (LLMs) and the building of a product platform with worldwide reach. Now, about two years later, third-party LLMs suffice much of the time, so the company shifting resources to post-training and “new product experiences.”
Character.AI has, therefore, signed a deal with Google, granting a non-exclusive license for its “current LLM technology” and gaining in exchange funding to expand and work on personalized AI. According to The Information, Google will buy out investors for about $88 per share, or about 2.5 times the price of Character.AI’s Series A shares.
A defection?
We learn from TechCrunch that much of Character.AI’s $150 million in funding came from the venture capital firm Andreessen Horowitz, also known as a16z, which, ironically, has an explicit “little tech agenda.” It was co-founded by Ben Horowitz and Marc Andreessen, both of whom have lately stepped into the political arena, in likely opposition to Google.
“We believe that advancing technology is critical for humanity’s future, so we will, for the first time, get involved with politics by supporting candidates who align with our vision and values specifically for technology,” Horowitz wrote in a blog post last December. “While ‘Big Tech’ is well represented in Washington D.C.,” he continued, “their interests are often at odds with a positive technological future as they are more interested in regulatory capture and preserving their monopolies.”
The firm, writes Horowitz, seeks an arrangement where AI operates “within the law and rules of our society” without any government regulation of “math, FLOPs, methods of R&D, and other misguided ideas.” Horowitz cites nuclear power as an industry that misguided regulation has choked off. “Decentralized technologies from the blockchain/crypto/web3 ecosystem will create a fairer, more inclusive economy than the centralized, monopoly-enabling technologies of web 2 and the industrial-era financial system.”
Potential regulatory scrutiny?
TechCrunch believes that regulators in the US and the EU might “scrutinize these reverse acqui-hires,” citing action by both the the UK Competition and Markets Authority (CMA) and the US Federal Trace Commission (FTC). The CMA issued a notice last month that it was opening a probe into Microsoft’s hiring of certain employees of another developer of personal AI, Inflection AI, and the “associated arrangements.” According to Investopedia, which cites the Wall Street Journal, the FTC opened a probe of its own about two months ago, because Microsoft had both paid $650 million to obtain a license for Inflection’s tech and hired most of the company’s staff.
