Law Firms Caught and Punished for Passing Around "Bogus" AI Slop in Court
As The Verge reported this week, the court filing in question was a brief for a civil lawsuit against the insurance giant State Farm. After its submission, a review of the brief found that it contained "bogus AI-generated research" that led to the inclusion of "numerous false, inaccurate, and misleading legal citations and quotations," as judge Michael Wilner wrote in a scathing ruling.
According to the ruling, it was only after the judge requested more information about the error-riddled brief that lawyers at the firms involved fessed up to using generative AI. And if he hadn't caught onto it, Milner cautioned, the AI slop could have made its way into an official judicial order.
"I read their brief, was persuaded (or at least intrigued) by the authorities that they cited, and looked up the decisions to learn more about them — only to find that they didn't exist," Milner wrote in his ruling. "That's scary."
"It almost led to the scarier outcome (from my perspective)," he added, "of including those bogus materials in a judicial order."
A lawyer at one of the firms involved with the ten-page brief, the Ellis George group, used Google's Gemini and a few other law-specific AI tools to draft an initial outline. That outline included many errors, but was passed along to the next law firm, K&L Gates, without any corrections. Incredibly, the second firm also failed to notice and correct the fabrications.
"No attorney or staff member at either firm apparently cite-checked or otherwise reviewed that research before filing the brief," Milner wrote in the ruling.
After the brief was submitted, a judicial review found that a staggering nine out of 27 legal citations included in the filing "were incorrect in some way," and "at least two of the authorities cited do not exist." Milner also found that quotes "attributed to the cited judicial opinions were phony and did not accurately represent those materials."
As for his decision to levy the hefty fines, Milner said the egregiousness of the failures, coupled with how compelling the AI's made-up responses were, necessitated "strong deterrence."
"Strong deterrence is needed," wrote Milner, "to make sure that lawyers don't respond to this easy shortcut."
More on lawyers and AI: Large Law Firm Sends Panicked Email as It Realizes Its Attorneys Have Been Using AI to Prepare Court Documents
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Forbes
3 hours ago
- Forbes
Google's NotebookLM Adds Video Summaries And Multi-Output Features
Google's NotebookLM has launched two significant updates : Video Overviews and an upgraded Studio panel that allows users to create multiple outputs from the same source material. NotebookLM, which Google calls a 'personalized AI research assistant,' is an AI model with some major differences between it and the major LLMs like ChatGPT, Claude, and Google's own Gemini. NotebookLM works with only the 'sources' you upload, not knowledge gained from the wider internet. This makes it more focuses and less prone to hallucinations. Although NotebookLM can generate a variety of content, it's been best known for its remarkably human-sounding, podcast-like Audio Overviews. Two realistic, engaging AI hosts discuss your content in a way that you can customize. Video Overviews: Visual Learning Made Simple When I heard that Video Overviews were coming, I thought Google might add human avatars to the Audio Overviews. That seemed fairly straightforward using technology like that of HeyGen or ElevenLabs. I'd simulated that concept with a bit of manual editing, and the result was fairly pleasing. The process was complicated only by the male and female host voices being intermingled in a single audio stream, a factor which wouldn't hamper Google's ability to add avatars. Instead, Google has taken a different direction for Video Overviews. Rather than a video podcast look with two hosts conversing on a split screen, the new feature creates a narrated slideshow. NotebookLM creates PowerPoint-like slides that combine AI-generated visuals with images, diagrams, quotes and numbers from the source documents. Users can customize these videos just as they do with Audio Overviews. You can specify topics to focus on, indicate your learning goals, describe the target audience and much more. The system handles both basic requests ("help me understand the key findings in the paper") and expert-level customization ("Explain the research methodology for an audience of behavioral scientists"). Tailored Content From the Same Source Material The upgraded Studio panel solves a significant limitation: until now, you could only create one of each per notebook, which could be limiting. If you used the sources to create a beginner's guide, you would lose that if you created a new overview tailored to experts. Now users can generate and store multiple outputs of the same type within a single notebook. This enables practical applications across organizations: The interface includes four tiles at the top of the Studio panel for creating Audio Overviews, Video Overviews, Mind Maps, and Reports with a single click. Users can now multitask within the panel—listening to an Audio Overview while reviewing a Mind Map, for example. Business Applications Worth Noting These features offer concrete benefits in a variety of situations. Marketing teams can upload competitive analyses and generate video summaries for stakeholder meetings. Product teams can transform technical specifications into accessible overviews. HR departments can create role-specific training materials from policy documents. My favorite use is turning lengthy, jargon-laden academic research into shorter, more digestible formats. The ability to generate multiple outputs lets one make complex information accessible to audiences with different expertise levels and learning preferences. Availability and Next Steps Video Overviews are now rolling out to all users in English, and support for more languages is coming soon. The redesigned Studio panel will reach all users over the coming weeks. Though less-well known than the major AI models, NotebookLM is a remarkably versatile tool. One personal example: I dropped the 350-page manual for my new car into a notebook. When I'm puzzled by how to change a setting buried in a third-level touchscreen menu, I ask NotebookLM and get an instant reference. That's much easier and faster than trying to locate the information in the enormous PDF from the car maker. Overall, I think Google's calling NotebookLM a 'personalized AI research assistant' is a bit of a misnomer. Where the tool shines isn't doing research. Rather, it is taking research you (or another AI) has gathered and extracting information and/or making the information highly accessible to a variety of audiences. The new features are another step in that direction.
Miami Herald
4 hours ago
- Miami Herald
Veteran analyst drops bold price target on Palantir stock ahead of earnings
Palantir (PLTR) stock's up an incredible 106% year to date, but that leaves investors wondering how much gas is left in the tank. Veteran Wall Street research firm Loop Capital just answered that question with a bold new target, and the logic goes beyond earnings beats. Don't miss the move: Subscribe to TheStreet's free daily newsletter Loop Capital sees something brewing under the surface, leaning into control, momentum, and a long-term edge in AI that no one's talking about. If anything, the bar just got higher, and Palantir's AI hype train isn't likely to slow down anytime soon. Image source: Bloomberg/Getty Images Palantir isn't looking to go toe-to-toe with ChatGPT, Google's Gemini, or Grok on base models. Its strength lies in what comes after, where operational layers turn those models into robust, usable, high-stakes tools. Related: Warren Buffett unloads $1.2 billion of this popular tech stock At the heart of it all is AIP, Palantir's potent Artificial Intelligence Platform. Built into its powerful Foundry software and closely integrated Gotham and Apollo platforms, AIP essentially gives clients a "control plane" for AI. That effectively gives them access to robust large language models (LLMs) and multimodal AI, bespoke agents, while constantly testing performance. This one-stop shop setup takes raw AI output into decision-grade actions. For enterprise and government clients, that's what transforms a base model from experimentation into a real-world deployment. Take the partnership with Microsoft, for instance. Palantir layers its comprehensive governance tools on top of Microsoft's cloud behemoth, even inside classified defense environments. That allows U.S. intelligence agencies to run sensitive AI workloads with full oversight efficiently. Then you have Palantir's forward-deployed engineers, who embed clients with customized AI workflows to accelerate adoption. Strategic alliances with Databricks and Accenture Federal Services help Palantir spread its tentacles even more, enabling real-time, autonomous decision-making in both commercial and federal settings. Moreover, with multi-hundred-million-dollar Department of Defense deals under its belt, Palantir's position as the "AI control layer" is essentially mission-critical. Palantir Technologies just got a massive price target bump. Loop Capital's Mark Schappel raised his price target to $178 from $155, reaffirming a Buy rating. The move comes just days ahead of Palantir's Q2 earnings report on August 4, which Schappel expects will be another quarter of outperformance in both sales and guidance for the AI software firm. Related: Morgan Stanley names tech stock winner from One Big Beautiful Bill Act Also, Schappel is modeling more upside in line with Palantir's recent beats, roughly 4.3% above guidance. He also expects the business to raise forward-looking numbers in the process. However, with the stock's blistering rally, where it's up 106% this year and 26% since Q1 earnings, it adds to the tension. However, Schappel believes it not only can keep going, but also may just be getting started. A big part of that is Palantir's multifold AI edge. It sits quietly at the intersection of AI and enterprise software, with powerful commercial tailwinds and a deep federal bench. This includes ties to Trump-aligned policy shifts linked with defense and digital modernization. AI adoption is expanding across multiple sectors, and Palantir is moving quickly in converting pilot deals into powerful production contracts. On top of that, the company posted a "Rule of 83" in Q1, a feat few companies in tech can match. To put things in perspective, the Rule of 83 is just its Rule of 40 score hitting 83%, which is essentially the sum of YOY revenue growth rate and adjusted operating margin (that sits well above the 40%). Veteran fund manager of TheStreet portfolio Chris Versace, though, has a different take on Palantir stock. He feels Palantir is bumping up against his price target, and recent trade developments and defense spending impacts have him revisiting his take. More News: Amazon's quiet pricing twist on tariffs stuns shoppersNvidia avoids White House crackdown; Trump softens on AI giantBank of America flags 3 breakout stocks to watch ahead of earnings He's already raised the panic point to $115 and set a pick-up at $137, which indicates that the stock is extended in an overbought market (considering it's trading at $158.61). Hence, it implies locking gains or waiting for a pullback before loading up more. Q2 earnings preview: Palantir must clear a high bar to keep rally alive Veteran investors are eyeing what could be a big step up heading into Palantir's Q2 print. The company had a standout Q1, where sales surged 39% YOY to $884 million, with U.S. revenue alone jumping 55% and U.S. commercial business growing 71%. That strength has led the management to bump its full-year revenue guidance to $3.89–$3.90 billion. Executives leaned hard into Palantir's powerful AI Platform and its durability amid scrutiny over U.S. defense spending. Consequently, Q2 expectations have climbed immensely. Mr. Market is now modeling $939.47 million in sales, a 38% YOY jump, alongside 56% EPS growth to 14 cents on a non-GAAP basis. For context, it posted a Q2 2024's beat of $678.13 million and 9 cents EPS. The implication is that Palantir must continue delivering superb results across commercial and federal pipelines, as it comes up against steep comps. Over the past 90 days, analysts have effectively revised earnings estimates 12 times higher and just three times lower, reflecting superb confidence. Those upgrades stem from stronger visibility into contract wins, scaling AI deployments, and healthy upside from federal defense initiatives. Related: Top economist drops 6-word verdict on Trump tariffs, inflation The Arena Media Brands, LLC THESTREET is a registered trademark of TheStreet, Inc.
The Verge
4 hours ago
- The Verge
Why AI researchers are getting paid like NBA All-Stars
Hello, and welcome to Decoder! This is Alex Heath, your Thursday episode guest host and deputy editor at The Verge. Today, I'm joined by Hayden Field, The Verge's senior AI reporter. We're talking about the AI talent wars and why some researchers are suddenly getting traded like they're NBA superstars. If you've been paying attention at all to the AI industry this past month, you've likely seen stories about the hiring frenzy that's been happening at Big Tech, research labs, and AI startups. Leading the charge is Mark Zuckerberg, who's been making some truly unprecedented job offers to try and lure top talent to Meta. But there's a whole lot more to this story. Hayden and I have been reporting on this for the past several weeks to get a sense of how big some of these offers really are, why companies like Google are opting to hire instead of acquire, and what it means that some of the most sought-after AI experts in the world are no longer motivated by money alone. If you'd like to read more on what we talked about in this episode, check out the links below: Questions or comments about this episode? Hit us up at decoder@ We really do read every email! A podcast from The Verge about big ideas and other problems. Posts from this author will be added to your daily email digest and your homepage feed. See All by Alex Heath Posts from this topic will be added to your daily email digest and your homepage feed. See All AI Posts from this topic will be added to your daily email digest and your homepage feed. See All Business Posts from this topic will be added to your daily email digest and your homepage feed. See All Decoder Posts from this topic will be added to your daily email digest and your homepage feed. See All Meta Posts from this topic will be added to your daily email digest and your homepage feed. See All OpenAI Posts from this topic will be added to your daily email digest and your homepage feed. See All Podcasts Posts from this topic will be added to your daily email digest and your homepage feed. See All Tech
