ANTHROPIC: PUNISHED FOR BEING TOO GOOD AT WHAT IT DOES?
As somebody who works with lots of AI startups (as well as other businesses that use AI), I’m monitoring this Anthropic situation with great interest.
In case you haven’t seen it yet, Anthropic recently released its two newest models, Fable 5 and Mythos 5, to glowing reviews from techies. And yet, within just a couple of days, the US government has forced them to take the product off the market.
The stated reasons are, if accurate, definitely serious. Anthropic thinks the US government believes a method has been identified to bypass a critical safeguard which otherwise prevents bad actors from easily using Fable 5’s impressive capabilities to hack into sensitive, very high-importance computer systems in the US and abroad. (Anthropic doesn't deny the severity of the hypothetical risk, but it does strongly challenge whether the risk is real in this case.)
Anyway, because of this supposed risk, the US government declared that all “foreign nationals” were immediately banned from accessing either Fable 5 or Mythos 5. This, in turn, required Anthropic to immediately disable these new models for all its users to be compliant with the government order.
The matter is further complicated by the fact that Anthropic has already incurred the ire of the US government when it refused to make its AI platform available to the Department of War for mass surveillance usage.
Even more interesting, this federal government intervention has come right before Anthropic launches its IPO. That's not to say the security concerns aren’t valid, but the optics around the timing make this move by the US government seem almost punitive in nature.
On top of all that, according to the second source I’ve linked below, the “whistleblower” that revealed this security concern to the administration is none other than the CEO of Amazon. Even though Amazon is a major stakeholder in Anthropic.
My takeaways here are nothing super unusual (and I am certainly not the first person to suggest any of them), but, if I had to leave you with something:
Companies (especially tech companies) that want to use the bleeding edge of AI to power their growth should probably have arrangements with multiple platforms in place. Because if it could happen to Anthropic, one of the AI "heavyweights", it's easy to see how this could happen to other major and minor AI platforms.
Keep an eye out for legal issues in this space in general. The AI field is still chaotic and fast moving. Who knows what will happen next? I, for one, am very curious what sort of lawsuits, if any, will come out of this ruling.
Beyond all of that, this news story is ironic to me because it seems that Anthropic is, in a way, being punished for being *too good* at what it does. And that is definitely not the usual source of legal liability when it comes to the tech space.
ASK A QUESTION OR SCHEDULE A MEETING/CALL.
Disclaimer: This article constitutes attorney advertising. Prior results do not guarantee a similar outcome. MGLS publishes this article for information purposes only. Nothing within is intended as legal advice.