Avis Short Sellers Caught In A Vice; Spirit Airlines Look To Trump For Salvation; and Anthropic’s Cybersecurity Model Has Been Hacked
- Dipo Owolabi
- Apr 23
- 5 min read
Updated: Apr 25
Markets and AI are colliding in ways that are hard to ignore. Avis Budget has become the latest symbol of how violent a short squeeze can be when bearish bets get crowded, while the Trump administration is nearing a bailout-style rescue for Spirit Airlines as fuel costs spike. At the same time, Anthropic’s high-risk cybersecurity model Mythos appears to have leaked beyond its intended circle, and Florida’s attorney general has escalated his inquiry into OpenAI into a criminal investigation. From trading chaos to AI power struggles and legal exposure, all this and more in today’s Read It And Eat! |
Markets Around The World
Markets as of 22nd April 2026.. Cells in RED mean that the value is down, cells in Green mean the value is up.
MAJOR HEADLINES
Avis Shares Skyrocket As Short Sellers Get Choked Out
Avis Budget extended its April rally to a record high as a brutal short squeeze punished investors betting against the stock. Reuters reported that as of Monday, 86.2% of Avis’ free float was shorted, only slightly below the 89.3% all-time peak recorded in March, making it one of the most crowded bearish trades in the market. The shares’ violent move higher forced shorts to unwind positions, accelerating the rally rather than stopping it.
What makes the move even more dramatic is how concentrated the ownership has become. Reuters said hedge funds SRS Investment Management and Pentwater Capital Management together controlled more than 25.2 million shares, equal to over 71% of total outstanding shares, while Pentwater’s increasing stake shrank the available float even further. That kind of ownership structure can make a stock extremely sensitive to supply and demand imbalances, especially when short sellers are already heavily exposed.
The losses on the short side were already enormous. Reuters said short sellers had lost about $4.09 billion so far in April, including nearly $1.01 billion on Monday alone, while trading was halted multiple times during Tuesday’s session. Avis had also reported weaker fundamentals earlier in the year, including a quarterly revenue decline and a $856 million net loss for the December quarter, which shows that the stock’s explosive move is being driven by positioning rather than a clean turnaround story. Reuters
Spirit Airlines Seek Bailout From Trump Administration
The Trump administration is nearing a rescue package for Spirit Airlines that could total as much as $500 million in government-backed financing. Reuters reported that the likely structure would be a loan to keep Spirit operating through bankruptcy, with the financing later turning into a longer-term loan when the airline exits bankruptcy. In exchange, the U.S. government could receive warrants that give it a potential stake of up to 90% in the carrier.
The urgency comes from the fuel shock hitting the airline industry after the Iran conflict disrupted oil flows through the Strait of Hormuz. Reuters said jet fuel prices have roughly doubled, and Spirit’s own turnaround plan assumed fuel of about $2.24 a gallon in 2026 and $2.14 in 2027; by mid-April, prices were around $4.24 a gallon. That gap has seriously weakened the airline’s restructuring plan and made its already fragile balance sheet even harder to manage.
Spirit is especially exposed because it has been shrinking its fleet and trying to survive on tighter margins. Reuters said the airline plans to cut down to roughly 76 to 80 aircraft by the third quarter of 2026, about a third of its pre-bankruptcy size, while officials and analysts warn that a bailout could create a precedent for rival carriers to seek similar help. The broader issue is that if Spirit fails, fares could rise in key markets, but if it gets a rescue, Washington risks being pulled deeper into the business of keeping airlines alive. Wall Street Journal
Anthropic’s newest cybersecurity-focused model, Mythos, is already under pressure after a report that unauthorized users gained access to it through a third-party vendor environment. TechCrunch said the company is investigating the claim and, so far, has found no evidence that its own systems were affected. The model is not a consumer product; Anthropic positioned it as an enterprise security tool with the potential to become a powerful hacking instrument if misused.
The access apparently came through an outside contractor rather than a direct compromise of Anthropic itself. TechCrunch reported that a private online forum managed to reach the tool, and that the group used multiple strategies to find it, including exploiting knowledge linked to a third-party worker’s access. Bloomberg’s reporting, as summarized by TechCrunch, said the users also produced screenshots and a live demo to prove they had reached the model.
That is exactly why Mythos has become such a sensitive product. Anthropic had already warned that the model could be dangerous in the wrong hands because it is capable of finding vulnerabilities and exploiting them. A security tool with that level of power is inherently double-edged: it can help defenders test systems, but it can also become a weapon if access controls fail. The current incident underscores how difficult it is to contain frontier AI once it is shared, even narrowly, with third-party vendors. Techcrunch |
Florida Begins Criminal Investigation Into OpenAI
Florida’s attorney general has escalated the state’s inquiry into OpenAI into a criminal investigation, issuing subpoenas for information about how the company handles users who threaten harm to themselves or others. NBC News reported that the office is specifically looking at whether ChatGPT offered “significant advice” to the suspect in the Florida State University shooting, which killed two people and injured five others.
The case is centered on disturbing chat records. According to NBC, investigators found that the suspect asked ChatGPT questions about the best gun, ammunition, and the optimal time and place on campus to kill the most people, and that the system responded. The attorney general said prosecutors would treat a human giving that kind of guidance as a murder charge, which is why the state is now demanding internal OpenAI policies and training materials going back to March 2024.
The broader significance is that AI safety is no longer just a policy debate; it is becoming a law-enforcement issue. OpenAI is already under pressure from regulators and critics over harmful outputs, but this case raises the stakes because it ties the product directly to a real-world violent event. If Florida’s probe advances, the outcome could shape not just OpenAI’s legal exposure, but how AI companies are expected to monitor threats, report dangerous behavior, and respond when their systems are used in life-or-death situations. NBC News
Minor Headlines
|
Comments