Microsoft has developed an AI assistant for lawyers within Word. Let's hope it doesn't act unpredictably.

      Digital Trends might receive a commission if you make a purchase through links on our website. Why should you place your trust in us?

      Microsoft has introduced an AI legal agent designed to review contracts and address trust issues.

      Microsoft Word will be equipped with an AI legal agent, which seems beneficial until recalling past failures in this area. The new Legal Agent can assess contracts, suggest modifications, compare versions, and identify risky clauses within Word. While these features appear to be quite helpful, there have been instances where generative AI tools have fabricated entire cases, citations, and quotes, resulting in real legal problems for individuals.

      What functions does Microsoft’s Legal Agent have?

      According to Microsoft, Legal Agent is accessible via Copilot in Word for users in its Frontier program in the U.S. Currently, it operates on Word for Windows desktop. No separate application or installation is needed, though some users might have to restart Word for the agent to activate.

      Legal Agent is specifically aimed at reviewing contracts and documents. Microsoft claims it can examine contract clauses one by one against a legal playbook, analyze an entire agreement, compare various versions, highlight risks and obligations, and suggest edits with tracked modifications. It also preserves original formatting, tables, lists, and the history of negotiations.

      The company is also aiming to prevent potential negative scenarios for both users and itself. The feature includes safeguards, such as providing citations linked to the source text so reviewers can verify suggestions before relying on them, alongside clear disclaimers indicating it does not offer legal advice, might generate inaccurate content, and still necessitates review by a qualified legal expert before anything can be trusted.

      Why should legal professionals still feel concerned?

      There are existing precedents for AI misfires in legal contexts, as evidenced by two New York attorneys receiving sanctions in 2023 and being fined $5,000 for submitting a court document that included fictitious cases created by ChatGPT. Michael Cohen, former lawyer for Donald Trump, also acknowledged he unknowingly provided his attorney with fake case citations generated by Google Bard. While Cohen faced no sanctions, the judge deemed the situation embarrassing and emphasized the importance of skepticism when utilizing AI in legal matters.

      These instances are not isolated; judges have questioned or penalized attorneys on multiple occasions regarding AI-assisted submissions, and one French data scientist and lawyer uncovered hundreds of court documents containing false citations and nonexistent references over the past year.

      The most significant issue is that hallucinations in AI remain unresolved. AI chatbots can deliver answers that sound confident yet are partially or entirely incorrect. In the legal field, this poses significant risks, as a fabricated citation or nonexistent case may make its way into a filing and lead to serious repercussions.

      Microsoft has implemented numerous safeguards within Legal Agent to mitigate these problems; however, the lessons learned are already recorded in legal histories. AI can expedite legal tasks, but the onus of fact-checking remains with the lawyer.

      I have four years of experience primarily focusing on gaming, PC hardware, and smartphones. In my leisure time, I enjoy...

      I allowed Gemini to manage my Gmail, and it was an enlightening experience.

      My inbox often resembles chaos, filled with everything from meeting requests and marketing pitches to important updates and an ongoing influx of urgent items. It can quickly become overwhelming. There are times when I avoid checking emails altogether because it feels like too much to handle, and I constantly worry I might miss something critical lost in the noise. This is precisely where Gemini made a difference for me. Its integration into my inbox feels like a safety net, assisting me in managing the clutter without the constant pressure to catch up.

      US tech companies are laying off staff to invest in AI; China considers this illegal within its borders.

      There’s a unique cruelty to Zhou’s situation that I can’t shake off. He spent his workdays interacting with AI—testing, correcting, and improving it—only to see that same technology give his employer justification to dismiss him. His company, a tech firm in Hangzhou, replaced him with the large language models he was hired to oversee, offered him a lesser position with a 40% salary reduction, and terminated his contract when he refused to accept it. A court has now ruled this illegal on two occasions. What US companies are pursuing openly, Chinese courts are now preventing.

      Space-based data centers may seem like a far-fetched idea. What if we place them on lamp posts instead?

      Nigeria has just become home to Africa's first distributed AI data center, integrated into 50,000 lamp posts.

      SpaceX has ambitious plans for AI data centers in space, while Microsoft has considered underwater installations. However, constructing AI data centers is both costly and power-hungry. This is why a UK company aims to create one using street lamp posts in Nigeria and has already signed an agreement to move forward. Warwickshire-based Conflow Power Group has partnered with Nigeria's Katsina State Government to deploy 50,000 solar-powered smart lamp posts, known as iLamps, across the state (via