Are we the AI baddies, again?

Until recently, Anthropic’s played the game well. It poo-pooed government contracts that might see its tech be used in untoward ways. It begged regulators to move in and slow down frontier development, lest they create a monster. And it flagged the capacity of Mythos to disrupt cybersecurity. Its chief competitor has generally done the opposite.

Deciding who’s a goodie and a baddie in a world of complex ethical decisions is a fruitless task, though—as demonstrated by Anthropic’s recent readmission to the naughty list. Public sentiment towards the company has cooled, largely because Amodei has refused to sign a statement supporting open-weight models, moving them to the wrong side of an increasingly heated political debate. Moreover, Anthropic has recently doubled down on anti-open-weight and open-source LLMs.

But that’s not the only naughtiness the company has been accused of. Indexed chat logs are increasingly available online, in most cases meaning anyone who had shared a chat within Claude would now likely see the contents all over Google. A spokesperson from Anthropic said “When someone shares a conversation, they are making that content publicly accessible, and like other public web content, it may be archived by third-party services.” Judging from the uproar across forums, including those we track here at Short Fuse Research, few knew this would be the case.

They’ve also ruffled the feathers of partners by launching products that stomp aggressively on their playing field. Figma is pissed about Claude Design, for example. Following the launch, its CEO argued that Anthropic had not been “consistently candid in their communications”. A chorus of leaders of other companies watching from the sidelines suggest Anthropic stitching up a partner has given them pause on whether Anthropic is the right partner for them. And it’s fair to say the large ecosystem of AI wrappers providing thin veneers of value over the top of Claude are now starting to worry that they’re next.

The result? A potential reversal of our bet that Anthropic could farm goodwill to win market share—though in a market where AI is rapidly becoming an unfeeling commodity, the whole “goodie vs. baddie” debate might just be noise to users who ultimately couldn’t care less.