LSU nearly got themselves kicked out of the SEC in a legal standoff over signing former NFL players to the roster. The SEC’s argument rested, in part, on the claim that LSU had no institutional control over its big money athletic coaches. And now we hear that LSU almost ran itself over the cliff because its head football coach kept taking legal advice from ChatGPT.
Gotta hand it to the SEC on this one.
ESPN just published a deep dive into LSU’s flirtation with athletic disaster. Coach Lane Kiffin — having spent last year living out a contracts issue-spotter exam — attempted to sign former Ole Miss players Dae’Quan Wright and Zxavian Harris, both of whom had signed NFL contracts, to play for LSU this year. The impetus was an NCAA ruling allowing an extra year of eligibility, which Kiffin read as allowing him to clawback players who already went pro. A Louisiana state judge saw no problem with this. The SEC responded with a threatened fine equal to half the football budget plus a half-season suspension for the coach, sued LSU in federal court in Alabama, and scheduled a vote of the conference presidents on whether to boot the university from the league entirely.
After a week of Louisiana politicians lining up behind Kiffin and loudly taunting the SEC, the school eventually did its research and backed down.
But here’s the legal news nugget from the ESPN story that struck us. Tom Mars — a Kiffin friend who made his name beating the NCAA on transfer waivers — wasn’t representing LSU or Kiffin in the case, but was weighing in from the sidelines, telling his friend that he really needed to back down. Here’s how that went:
“Usually, when I would express an opinion about what LSU was doing, Lane would reply with a screenshot from ChatGPT that supported LSU’s position, which was almost always wrong,” Mars said. “Lane ignored most of my unsolicited advice, but I don’t think he’ll be using ChatGPT anymore for legal advice.”
Eventually, Mars told him that “a guy with a $91 million contract ought to have a top-notch lawyer on speed dial and not be using ChatGPT for legal advice.”
In Kiffin’s defense, as long as he kept the school in state court, ChatGPT was doing just fine. The problem was that the other universities in the SEC were ready to drag this into federal court. But this is the reality of legal tech right now — litigants are armchair quarterbacking their strategy with help from free bots as real lawyers scream into their pillows.
Lane Kiffin may be the greatest possible spokesperson for generative AI. For all its advancements, the technology remains, fundamentally, “Mansplaining as a Service” — a fluent, confident explanation of anything, on demand, with no particular relationship to whether it’s right. Lane Kiffin’s entire career of failing upward is the football corollary of someone texting a chatbot’s legal analysis at an actual lawyer. When Mars says the screenshots Kiffin sent “supported LSU’s position,” it’s a tell. The bot wasn’t parsing the NCAA manual, it took LSU’s self-interested prompt and reverse engineered an answer.
Most users don’t understand that, and it’s what makes the technology dangerous. Lawyers told Axios that ChatGPT is telling would-be clients “You got a killer case” — “the WebMD effect on steroids,” as one put it. The professionals aren’t immune either. Damien Charlotin’s database of AI hallucination cases sits at 2,041 and counting.
ChatGPT aside, the fact that LSU had its football coach looking up legal strategy with a chatbot and forwarding it to a lawyer who didn’t even represent the school smacks of “lack of institutional control.” But LSU ultimately gave up on the NFL players so the SEC decided it wasn’t a problem anymore.
Legal tech products exist to prevent exactly this. The whole industry exists to enrich context and throw up guardrails to make AI deliver good results. Which is all well and good for the lawyers, but that’s not what the masses are seeing. They’re typing leading questions into free sycophant machines and basing their legal decisions off the output.
Despite the fever dreams of the people bankrolling the AI industry, the public doesn’t want to pay for AI at all. Millions of people have AI accounts and most will say they think AI is kind of cool, but when you hunker down and look at the numbers, only 3 percent of U.S. households pay for AI for personal use. ChatGPT has around 900 million weekly users and about 5 percent of them pay. People want to ask a free bot whether their landlord can keep the security deposit, maybe twice a year, and then close the tab.
This is the folly driving the American AI project. The people building it think they’re assembling a mechanical messiah. Achieving “AGI” will, to their mind, usher in a science fiction utopia. The investors pumping money into the labs think AGI will allow them to fire every worker in the economy and convert the poor into the human battery pods from The Matrix. Different delusions, but if you buy either, you genuinely believe that no amount of money is too much to waste in pursuit of that goal.
The problem is they’re never going to achieve it and the public doesn’t care. At the end of the day, LLMs are just word calculators. Well, technically all calculators are word calculators, but these produce more than “80085.” For most users, that added vocabulary is interesting, but ultimately not worth paying for.
For enterprise users, who actually get value out of AI, that means Tokengeddon is nigh. When AI labs need to start showing real revenue, the business users will have to start paying for what this stuff is actually worth. Because the consumer user already refuses to pay anything. When that bill comes due, everyone’s going to start looking around to see if there’s a cheaper alternative.
And it turns out… there are cheaper alternatives! Open weight models, especially those coming out of China, have taken more than 30 percent of weekly token consumption on OpenRouter since February, peaking at 46 percent, while running 60 to 90 percent cheaper than the American alternatives. Legal tech is right there with them. Harvey — a company OpenAI invested in, by the way — built an in-house legal model based off Moonshot’s Kimi K3. Thomson Reuters built its model drawing on Alibaba’s Qwen (though CoCounsel still runs mostly on Claude).
About a year ago, I remember telling people at a legal tech conference that whoever invents an “American DeepSeek” — a smaller, cheaper model that can admit what happened at Tiananmen Square — would end up winning the AI race. It turns out the market isn’t willing to wait for the “American” part.
Which brings us to this week’s sudden obsession with AI caution. On September 12, Anthropic CEO Dario Amodei published “We Must Pace the Frontier,” declaring that “we must slow the pace at which we improve the capabilities of AI models.” Sam Altman chimed in that “I agree with Dario that we need to pace the frontier.” Elon Musk: “Dario is right.” Social media’s preeminent corporate mouthpieces lit up claiming that everyone was sleeping on the threat of AI and the time for regulation had arrived.
The plan asks frontier companies to coordinate on “limits on the rate of unchecked AI progress,” and Amodei knows how that sounds:
For antitrust reasons, it’s helpful for the US government to mediate or at least enable these discussions — they don’t need to participate, but do need to issue a narrow waiver for certain kinds of safety conversations.
Fun fact: there’s no antitrust barrier to slowing down and there’s no need for new laws to govern AI bots committing crimes. Existing laws cover all of this right now. What the AI companies want from the government isn’t regulation, it’s a new regime that explicitly grants them legal protections while throwing entry barriers in front of future competitors. And if they can protect the industry’s circular financing of massive physical data centers — enormous capital expenditures that become meaningless when people start installing small models on their laptops — all the better. We’re going to hear horror stories about rogue AI and the existential risk of Skynet becoming self-aware at 2:14 a.m., EDT on August 29, followed immediately by the frontier labs explaining that ONLY WE can prevent these dangers.
Remember when this guy said Biglaw had replaced junior associates with AI? Weird how he keeps having these private conversations where he keeps hearing the industry’s scariest public relations pitch. But, yeah, we’re going to see a lot of stories like these.
And then a push for some new law that manages to make OpenAI, Anthropic, Google, and maybe Elon the only “safe” providers.
Anthropic says it “has never advocated for a ban on open-weights models” and that “open-weights models that don’t have dangerous capabilities are a public good.” However, the essay calibrates this whole slowdown to America’s lead, warning that any slower and “(unpaced) CCP-associated projects will pull ahead.”
What does this mean? No one knows!
AI is an existential threat and also a race with China. Whatever… the point is, everyone needs to be really scared and give the AI providers exactly what they want.
Which somehow brings us back to the SEC and ESPN.
On the same Saturday Amodei published his manifesto, ESPN gave Ted Cruz roughly 10 minutes of College GameDay in Austin to pitch the Protect College Sports Act while Texas students chanted “Ted, you suck!” — the sharpest legal analysis of the morning. The whole thing was such a disaster, you expected Cruz to immediately leave for Cancun. In reality, ESPN lobbied for the bill, designed to reclaim power from student-athletes to the benefit of the schools and, importantly, the broadcasters. If the ploy were any more transparent, it would be an Oregon offensive lineman.
Three days later, the Senate voted 74-24 to advance it.
The Protect College Sports Act hands the NCAA and conferences immunity from federal and state antitrust law to enforce eligibility, transfer, and compensation rules. An amendment to the Sports Broadcasting Act allows conferences to pool and sell media rights like pro leagues. It bans outside entities from poaching schools into a Super League. Conference realignment is frozen to satisfy old men who can’t get over the Big 8 no longer existing. There’s even a coaching provision widely dubbed the “Lane Kiffin Rule,” to prevent someone copying Kiffin’s move of bolting Ole Miss on the eve of the playoff. Advocates for the bill say it’s about “saving the sport” and “helping the non-revenue athletes.” Meanwhile, everyone agrees that it’s still cool to pay coaches tens of millions of dollars not to coach.
So it’s a legal shield coupled with entry barriers on new competitors. Does any of this sound familiar?
And why are we here? Because ESPN threw its lot in with the SEC and the SEC is mad that everyone (else) can pay players now. For all the “it just means more” hype, the SEC wasn’t appreciably better than any other league for most of my young life. Miami, Nebraska, and Florida State combined for roughly 10 national titles between 1983 and 2001. We let BYU and Washington win titles in that stretch! In the early aughts, the SEC produced some dominant teams and locked in a 15-year, $2.25 billion deal with ESPN. A Sports Illustrated headline at the time explained: “De facto TV network will push SEC even further ahead of competitors.”
Now, the rules have changed and the SEC hasn’t won a title in three years. We’re letting Indiana win at football now. And this has turned College GameDay into Ted Cruz and Nick Saban complaining that players having power over their own careers has destroyed the sport while the ratings have never been higher.
But it’s not just the SEC! The Big Ten fought for this bill too. Think of them as the Anthropic to the SEC’s OpenAI. Now that they’ve secured their seat at the big table, they also want new rules to slow it down. The Big Ten built itself into a juggernaut by destroying the Pac-12 and fielding teams spraying a money cannons at players. Now they want brakes on conference realignment and paying players.
It’s all fun and games until you’re on top and then everyone wants the government to build them a moat. And they sell that moat with worrying stories about agents going rogue and hacking into competitors to cheat the system.
And I was talking about Michigan right there, but OpenAI does it too. Even though we have a law designed specifically to stop people from hacking into computer systems, OpenAI throws up its hands and declares that it can’t be held accountable because they didn’t intentionally do it. Sure they recklessly designed a system, set it up to hack, pointed it — with ultimately weak guardrails — at HuggingFace, and then it went ahead and did it, but they didn’t mean to.
When AI labs say we need new laws, they aren’t talking about adopting a recklessness standard for the CFAA, they mean a whole new system that enshrines their lack of liability. Just wait to see what they start lobbying for. If you think it’ll involve tough penalties for their behavior, you the kind of person who dumped your 401K into NFTs.
Anyway, this is a long strange trip to come back to Lane Kiffin giving the conference the precautionary example they craved. Look, we need this legislation or Lane Kiffin’s going to win the race and sign NFL players!
The fact that he did it with help from ChatGPT just adds a dash of irony.
Joe Patrice is a senior editor at Above the Law and co-host of Thinking Like A Lawyer. Feel free to email any tips, questions, or comments. Follow him on Twitter or Bluesky if you’re interested in law, politics, and a healthy dose of college sports news.
The post Lane Kiffin Is EXACTLY The Guy Who Uses ChatGPT For Legal Advice appeared first on Above the Law.
LSU nearly got themselves kicked out of the SEC in a legal standoff over signing former NFL players to the roster. The SEC’s argument rested, in part, on the claim that LSU had no institutional control over its big money athletic coaches. And now we hear that LSU almost ran itself over the cliff because its head football coach kept taking legal advice from ChatGPT.
Gotta hand it to the SEC on this one.
ESPN just published a deep dive into LSU’s flirtation with athletic disaster. Coach Lane Kiffin — having spent last year living out a contracts issue-spotter exam — attempted to sign former Ole Miss players Dae’Quan Wright and Zxavian Harris, both of whom had signed NFL contracts, to play for LSU this year. The impetus was an NCAA ruling allowing an extra year of eligibility, which Kiffin read as allowing him to clawback players who already went pro. A Louisiana state judge saw no problem with this. The SEC responded with a threatened fine equal to half the football budget plus a half-season suspension for the coach, sued LSU in federal court in Alabama, and scheduled a vote of the conference presidents on whether to boot the university from the league entirely.
After a week of Louisiana politicians lining up behind Kiffin and loudly taunting the SEC, the school eventually did its research and backed down.
But here’s the legal news nugget from the ESPN story that struck us. Tom Mars — a Kiffin friend who made his name beating the NCAA on transfer waivers — wasn’t representing LSU or Kiffin in the case, but was weighing in from the sidelines, telling his friend that he really needed to back down. Here’s how that went:
“Usually, when I would express an opinion about what LSU was doing, Lane would reply with a screenshot from ChatGPT that supported LSU’s position, which was almost always wrong,” Mars said. “Lane ignored most of my unsolicited advice, but I don’t think he’ll be using ChatGPT anymore for legal advice.”
Eventually, Mars told him that “a guy with a $91 million contract ought to have a top-notch lawyer on speed dial and not be using ChatGPT for legal advice.”
In Kiffin’s defense, as long as he kept the school in state court, ChatGPT was doing just fine. The problem was that the other universities in the SEC were ready to drag this into federal court. But this is the reality of legal tech right now — litigants are armchair quarterbacking their strategy with help from free bots as real lawyers scream into their pillows.
Lane Kiffin may be the greatest possible spokesperson for generative AI. For all its advancements, the technology remains, fundamentally, “Mansplaining as a Service” — a fluent, confident explanation of anything, on demand, with no particular relationship to whether it’s right. Lane Kiffin’s entire career of failing upward is the football corollary of someone texting a chatbot’s legal analysis at an actual lawyer. When Mars says the screenshots Kiffin sent “supported LSU’s position,” it’s a tell. The bot wasn’t parsing the NCAA manual, it took LSU’s self-interested prompt and reverse engineered an answer.
Most users don’t understand that, and it’s what makes the technology dangerous. Lawyers told Axios that ChatGPT is telling would-be clients “You got a killer case” — “the WebMD effect on steroids,” as one put it. The professionals aren’t immune either. Damien Charlotin’s database of AI hallucination cases sits at 2,041 and counting.
ChatGPT aside, the fact that LSU had its football coach looking up legal strategy with a chatbot and forwarding it to a lawyer who didn’t even represent the school smacks of “lack of institutional control.” But LSU ultimately gave up on the NFL players so the SEC decided it wasn’t a problem anymore.
Legal tech products exist to prevent exactly this. The whole industry exists to enrich context and throw up guardrails to make AI deliver good results. Which is all well and good for the lawyers, but that’s not what the masses are seeing. They’re typing leading questions into free sycophant machines and basing their legal decisions off the output.
Despite the fever dreams of the people bankrolling the AI industry, the public doesn’t want to pay for AI at all. Millions of people have AI accounts and most will say they think AI is kind of cool, but when you hunker down and look at the numbers, only 3 percent of U.S. households pay for AI for personal use. ChatGPT has around 900 million weekly users and about 5 percent of them pay. People want to ask a free bot whether their landlord can keep the security deposit, maybe twice a year, and then close the tab.
This is the folly driving the American AI project. The people building it think they’re assembling a mechanical messiah. Achieving “AGI” will, to their mind, usher in a science fiction utopia. The investors pumping money into the labs think AGI will allow them to fire every worker in the economy and convert the poor into the human battery pods from The Matrix. Different delusions, but if you buy either, you genuinely believe that no amount of money is too much to waste in pursuit of that goal.
The problem is they’re never going to achieve it and the public doesn’t care. At the end of the day, LLMs are just word calculators. Well, technically all calculators are word calculators, but these produce more than “80085.” For most users, that added vocabulary is interesting, but ultimately not worth paying for.
For enterprise users, who actually get value out of AI, that means Tokengeddon is nigh. When AI labs need to start showing real revenue, the business users will have to start paying for what this stuff is actually worth. Because the consumer user already refuses to pay anything. When that bill comes due, everyone’s going to start looking around to see if there’s a cheaper alternative.
And it turns out… there are cheaper alternatives! Open weight models, especially those coming out of China, have taken more than 30 percent of weekly token consumption on OpenRouter since February, peaking at 46 percent, while running 60 to 90 percent cheaper than the American alternatives. Legal tech is right there with them. Harvey — a company OpenAI invested in, by the way — built an in-house legal model based off Moonshot’s Kimi K3. Thomson Reuters built its model drawing on Alibaba’s Qwen (though CoCounsel still runs mostly on Claude).
About a year ago, I remember telling people at a legal tech conference that whoever invents an “American DeepSeek” — a smaller, cheaper model that can admit what happened at Tiananmen Square — would end up winning the AI race. It turns out the market isn’t willing to wait for the “American” part.
Which brings us to this week’s sudden obsession with AI caution. On September 12, Anthropic CEO Dario Amodei published “We Must Pace the Frontier,” declaring that “we must slow the pace at which we improve the capabilities of AI models.” Sam Altman chimed in that “I agree with Dario that we need to pace the frontier.” Elon Musk: “Dario is right.” Social media’s preeminent corporate mouthpieces lit up claiming that everyone was sleeping on the threat of AI and the time for regulation had arrived.
The plan asks frontier companies to coordinate on “limits on the rate of unchecked AI progress,” and Amodei knows how that sounds:
For antitrust reasons, it’s helpful for the US government to mediate or at least enable these discussions — they don’t need to participate, but do need to issue a narrow waiver for certain kinds of safety conversations.
Fun fact: there’s no antitrust barrier to slowing down and there’s no need for new laws to govern AI bots committing crimes. Existing laws cover all of this right now. What the AI companies want from the government isn’t regulation, it’s a new regime that explicitly grants them legal protections while throwing entry barriers in front of future competitors. And if they can protect the industry’s circular financing of massive physical data centers — enormous capital expenditures that become meaningless when people start installing small models on their laptops — all the better. We’re going to hear horror stories about rogue AI and the existential risk of Skynet becoming self-aware at 2:14 a.m., EDT on August 29, followed immediately by the frontier labs explaining that ONLY WE can prevent these dangers.
Remember when this guy said Biglaw had replaced junior associates with AI? Weird how he keeps having these private conversations where he keeps hearing the industry’s scariest public relations pitch. But, yeah, we’re going to see a lot of stories like these.
And then a push for some new law that manages to make OpenAI, Anthropic, Google, and maybe Elon the only “safe” providers.
Anthropic says it “has never advocated for a ban on open-weights models” and that “open-weights models that don’t have dangerous capabilities are a public good.” However, the essay calibrates this whole slowdown to America’s lead, warning that any slower and “(unpaced) CCP-associated projects will pull ahead.”
What does this mean? No one knows!
AI is an existential threat and also a race with China. Whatever… the point is, everyone needs to be really scared and give the AI providers exactly what they want.
Which somehow brings us back to the SEC and ESPN.
On the same Saturday Amodei published his manifesto, ESPN gave Ted Cruz roughly 10 minutes of College GameDay in Austin to pitch the Protect College Sports Act while Texas students chanted “Ted, you suck!” — the sharpest legal analysis of the morning. The whole thing was such a disaster, you expected Cruz to immediately leave for Cancun. In reality, ESPN lobbied for the bill, designed to reclaim power from student-athletes to the benefit of the schools and, importantly, the broadcasters. If the ploy were any more transparent, it would be an Oregon offensive lineman.
Three days later, the Senate voted 74-24 to advance it.
The Protect College Sports Act hands the NCAA and conferences immunity from federal and state antitrust law to enforce eligibility, transfer, and compensation rules. An amendment to the Sports Broadcasting Act allows conferences to pool and sell media rights like pro leagues. It bans outside entities from poaching schools into a Super League. Conference realignment is frozen to satisfy old men who can’t get over the Big 8 no longer existing. There’s even a coaching provision widely dubbed the “Lane Kiffin Rule,” to prevent someone copying Kiffin’s move of bolting Ole Miss on the eve of the playoff. Advocates for the bill say it’s about “saving the sport” and “helping the non-revenue athletes.” Meanwhile, everyone agrees that it’s still cool to pay coaches tens of millions of dollars not to coach.
So it’s a legal shield coupled with entry barriers on new competitors. Does any of this sound familiar?
And why are we here? Because ESPN threw its lot in with the SEC and the SEC is mad that everyone (else) can pay players now. For all the “it just means more” hype, the SEC wasn’t appreciably better than any other league for most of my young life. Miami, Nebraska, and Florida State combined for roughly 10 national titles between 1983 and 2001. We let BYU and Washington win titles in that stretch! In the early aughts, the SEC produced some dominant teams and locked in a 15-year, $2.25 billion deal with ESPN. A Sports Illustrated headline at the time explained: “De facto TV network will push SEC even further ahead of competitors.”
Now, the rules have changed and the SEC hasn’t won a title in three years. We’re letting Indiana win at football now. And this has turned College GameDay into Ted Cruz and Nick Saban complaining that players having power over their own careers has destroyed the sport while the ratings have never been higher.
But it’s not just the SEC! The Big Ten fought for this bill too. Think of them as the Anthropic to the SEC’s OpenAI. Now that they’ve secured their seat at the big table, they also want new rules to slow it down. The Big Ten built itself into a juggernaut by destroying the Pac-12 and fielding teams spraying a money cannons at players. Now they want brakes on conference realignment and paying players.
It’s all fun and games until you’re on top and then everyone wants the government to build them a moat. And they sell that moat with worrying stories about agents going rogue and hacking into competitors to cheat the system.
And I was talking about Michigan right there, but OpenAI does it too. Even though we have a law designed specifically to stop people from hacking into computer systems, OpenAI throws up its hands and declares that it can’t be held accountable because they didn’t intentionally do it. Sure they recklessly designed a system, set it up to hack, pointed it — with ultimately weak guardrails — at HuggingFace, and then it went ahead and did it, but they didn’t mean to.
When AI labs say we need new laws, they aren’t talking about adopting a recklessness standard for the CFAA, they mean a whole new system that enshrines their lack of liability. Just wait to see what they start lobbying for. If you think it’ll involve tough penalties for their behavior, you the kind of person who dumped your 401K into NFTs.
Anyway, this is a long strange trip to come back to Lane Kiffin giving the conference the precautionary example they craved. Look, we need this legislation or Lane Kiffin’s going to win the race and sign NFL players!
The fact that he did it with help from ChatGPT just adds a dash of irony.
Joe Patrice is a senior editor at Above the Law and co-host of Thinking Like A Lawyer. Feel free to email any tips, questions, or comments. Follow him on Twitter or Bluesky if you’re interested in law, politics, and a healthy dose of college sports news.
