Corgi Launches AI Liability Insurance

Corgi, a new insurance company backed by Y Combinator, is now offering AI liability insurance – for both the AI companies providing the outputs, and the businesses – and potentially law firms – that use those AI tools.

AL asked CEO, Nico Laqua, about what was covered in terms of AI, especially whether it was both sides of the equation; and AL also noted that the legal world is very interested in this area.

‘Where the liability is, we cover’, he told this site on X.

And on the subject of legal AI, Laqua added: ‘Besides insurance, legal is the best use case of AI, IMO. Lots of productivity gains. I’m very excited about everything I’m seeing there and we are heavy users of AI native law firms.’

So, what are they offering? First, Corgi is a tech startup itself, as well as an insurance company. They’ve recently raised $108m. It’s also got a strong focus on tech startups in general, of all shapes and sizes. However, it’s their insurance coverage for AI issues in particular that is likely to get the most interest from the legal world.

It seems that what they offer is evolving rapidly – and yesterday’s announcement included mentions of insuring against agents going wrong. But, as far as AL can see at the moment, one policy that directly covers AI, that is listed on their website, is this one:

AI and Algorithmic Liability Endorsement (CORG-TECH-0038)

‘A modular endorsement for companies using AI, machine learning, or automated decision-making. You select the modules you need — each with its own limit and retention.

Available modules include:

• algorithmic bias liability,
• AI hallucination/defamation,
• training-data misuse,
• data poisoning/adversarial attacks,
• autonomous-AI bodily injury, deepfake and synthetic media,
• service interruption,
• AI intellectual property,
• regulatory investigation defense, and civil fines (where insurable).

For example: your AI lending engine is accused of discriminatory credit decisions (Algorithmic Bias module). [Endorsement required — coverage applies only to selected modules].’

Elsewhere the Corgi website notes that: ‘Tech & AI Liability is a specific evolution of professional liability insurance designed for artificial intelligence companies.

‘It covers legal defense and damages if a customer alleges that your AI model, software, or algorithm failed to perform as intended and caused them a financial loss. Corgi explicitly covers three AI risk categories: model performance & hallucination, algorithmic bias, and training data disputes.’

Although, it’s not clear if the immediately above connects to the policy AI and Algorithmic Liability Endorsement offering.

Overall, it feels like Corgi’s offering is rapidly expanding and evolving as new AI aspects come to their attention.

Important Note: this article is not insurance, legal, or financial advice. The AL news story here is solely an informational article.

–

Is this a big deal?

Yes. It’s been reported that some major insurance companies don’t want to get involved in AI-related coverage, in part because it’s such a new market space that they don’t know the risks, so want to avoid it.

Secondly, whenever AL’s founder gives a talk on legal AI, aside from ‘but how will we train junior lawyers in the future?’, one of the next most popular questions is ‘can you insure against what these tools do?’

As far as AL knows, there is only one legal AI company that specifically offers insurance against its own outputs, and that’s Orbital, in relation to residential property work.

So, this will be interesting to plenty of law firms, inhouse teams, and legal tech companies.

One final point: many have suggested that dependent on one’s local rules, anything a law firm sends to a client – no matter how it’s made – is the firm’s total responsibility. But, who knows how this position will evolve as ‘legal AI’ becomes synonymous with the outputs of law firms in general? In fact, some may argue we are already at that point. But, again, talk to your regulators and other experts!

You can find more about Corgi here.

—