Anthropic and Freshfields strike first-of-its-kind deal to build specialized legal AI tools

Anthropic and Freshfields agree deal to create legal AI tools

Freshfields has agreed a deal to help Anthropic build specialist legal AI tools that could later be sold to rival law firms, as the "magic circle" law firm seeks to increase its use of the technology. Under the deal, the maker of the Claude AI tool will draw on expertise from Freshfields' lawyers to build products that attorneys can use for tasks such as drafting documents, reviewing contracts and carrying out due diligence on companies. Freshfields will roll out Claude across its offices globally and gain early access to future Anthropic models and tools, though not the Mythos model whose ability to find vulnerabilities in cyber defences has sent shockwaves through global business. The agreement, the first of its kind that Anthropic has made with a law firm, deepens Freshfields' relationship with the fast-growing technology company, which is already a client. Anthropic, launched in 2021, has focused on developing tools for businesses. It raised $30bn from investors in February at a $350bn valuation and is preparing for a bumper initial public offering, possibly as soon as this year, the FT reported in February. "As we're all discovering the true potential of these tools . . . it's about combining our legal expertise with the technical expertise of Anthropic to try to come up with better products for ourselves and our clients," Freshfields' global managing partner Alan Mason told the FT. "We want to be leaders in this space because we believe it's absolutely essential as part of being [a] modern lawyer to use AI tools, the right tools in the right way," he said. Freshfields was running a "huge engagement programme" to encourage lawyers to make more use of AI, he added. Mark Pike, Anthropic's associate general counsel, said it was "so important for Anthropic to learn more about how modern law firms will work with in-house lawyers". He said the Freshfields tie-up was "a signal to every other law firm that moving slowly with AI deployment is not an option right now". He declined to share details of who would own the intellectual property of any new tools or whether Freshfields would receive revenue from future sales of any tools developed through the partnership. Under the agreement, Freshfields will pay Anthropic an undisclosed sum. Anthropic will not be able to use Freshfields' data to train its models. "The legal sector is complex, high stakes, and really has historically been resistant to new technology," Pike said. In-house lawyers were expecting the law firms that they used to "do more with less" and it "does feel like AI is going to lead the way in that transformation", he said. Law firms are making growing use of AI, but in some cases it has caused problems. Elite firm Sullivan & Cromwell apologised to a US federal judge this week because an important filing in a high-profile case contained multiple "hallucinations" by AI software, including misquoting the US bankruptcy code. Asked about the risks of hallucinations, Mason said Freshfields had a process in which "nothing goes out without being checked by individual lawyers". He said it was clear that in the future AI would "change the traditional law firm model, be it the delivery model, the pricing model, the cost model . . . but nobody yet has the full answer to what the future's going to look like". Freshfields last year struck a separate deal with Google to roll out its products, including AI tools, across the law firm and to use them to create bespoke AI agents for legal work. Anthropic has previously agreed partnerships in other industries, including with Accenture and Goldman Sachs. In the Goldman Sachs deal, the technology company embedded engineers within the bank to develop tools for accounting for trades and transactions, and client vetting and onboarding.

Anthropic, law firm Freshfields to jointly develop AI legal tools

WASHINGTON, April 23 (Reuters) - Global law firm Freshfields and artificial intelligence startup Anthropic said on Thursday that they signed an agreement to develop new AI tools, as large law firms look increasingly to embed generative AI into their legal services. London-based Freshfields will work with the legal team at Anthropic to develop AI applications for legal services, the firm said in a statement. The law firm, which employs more than 2,800 attorneys, will also receive early access to future Anthropic models and products as part of the deal. No financial terms were disclosed. Spokespeople for Anthropic and Freshfields did not immediately respond to requests ⁠for comment about their agreement. Large law firms have emerged as a key customer base for AI products, which can speed legal research and drafting of contracts and court documents. Two startups that develop AI products for lawyers, Legora and Harvey, have reached multibillion-dollar valuations in recent months. Reuters parent company Thomson Reuters also sells legal-focused AI tools, including CoCounsel which uses technology from Anthropic and other AI companies. Freshfields said on Thursday it is "an early adopter and tester" of CoCounsel's newest version, which Thomson Reuters said has been rebuilt to provide enhanced reasoning and reliability. In an email responding to a Reuters request for comment, a Thomson ⁠Reuters spokesperson said that Freshfields is a "valued partner" and an early innovator in applying AI to legal work. "The focus is now shifting from experimenting with AI to embedding it into day-to-day workflows at scale," Thomson Reuters Chief Technology Officer Joel Hron said in the email, describing CoCounsel as an example. Freshfields had already rolled out Anthropic's ⁠Claude chatbot models internally before Thursday's announcement, the firm said in its statement. The firm said it plans to expand to Anthropic's autonomous, or agentic, AI platform called Cowork. Anthropic released a legal plug-in for Cowork in January. Freshfields said it will ⁠work with Anthropic to create applications for legal and market research, contract review, document drafting and automation of internal business‑services workflows. AI's growing use for legal work has led to errors in court documents ⁠in some cases due to the technology's potential to "hallucinate" legal sources. Law firm Sullivan & Cromwell last week apologized to a federal bankruptcy judge after the firm submitted a filing with inaccurate citations and other errors generated by an unidentified AI program. Reporting by Mike Scarcella; Editing by David Bario Our Standards: The Thomson Reuters Trust Principles., opens new tab