8-minute read | 1,800 words
What to know this week
DC Circuit offers potential off-ramp to Anthropic-White House feud.
In the ongoing lawsuit between Anthropic and the White House, the federal appeals court overseeing the case could potentially provide a resolution to the months-long feud.
Poland passes new crypto regulations.
Poland has adopted a new bill that aims to better regulate cryptocurrencies amid a major cryptocurrency scandal.
This week's full stories
DC court could provide potential resolution to Anthropic’s lawsuit.
THE NEWS
On Tuesday, the DC federal appeals court overseeing Anthropic’s lawsuit against the White House could offer a potential resolution to the dispute. The trial is currently presided over by Judges Gregory Katsas, Neomi Rao, and Karen Henderson. Both Judge Katsas and Rao were appointed by President Trump with Judge Henderson being previously appointed by former President George H.W. Bush.
During the trial, Judges Katsas and Rao both voiced skepticism of the Court’s authority to review the Pentagon’s previous designation that Anthropic was a supply chain risk. However, the two did suggest that they could review specific orders that would ban a company and its products from government work.
Judge Rao commented that while the Court “can’t review designation decisions” it can emphasize that further directives excluding a company’s products from government use are unlawful. Additionally, the two judges suggested that Anthropic’s lawsuit may be premature after the Trump administration announced that it was reassessing the administration’s previous designation.
Outside of expressing concerns about jurisdiction, Judge Katsas also asked the Department of Justice attorney Sharon Swingle if the Trump administration would want an “informal” pause of the case to pursue a resolution with Anthropic.
THE KNOWLEDGE
After the Trump administration designated Anthropic as a supply chain risk and issued an Executive Order barring the use of their products within the federal government, the relationship between the two devolved as Anthropic filed two lawsuits. However, as these two lawsuits were ongoing, the dynamic between the two began to change with the introduction of Anthropic’s latest cybersecurity model, Claude Mythos.
Given the potential industry-changing impacts that Mythos could have, governments and industry leaders have been looking to get model access. This new model appears to have changed the Trump administration’s views on Anthropic and its newest model. At the end of April, reports began to emerge that the White House was considering different ways for the administration to be able to bring Anthropic back into the federal government while still “[saving] face.” Additionally, in early May, a Department of Defense technology official announced that the Pentagon was deploying Mythos cybersecurity model to find and patch software vulnerabilities across the US government.
This shifting policy stance is reflective of a broader shift taking place in the administration. Outside of improving relations between the White House and Anthropic, the administration is also changing its stance on AI oversight. Whereas previously the administration focused on a deregulated AI policy stance, Mythos and other similar advanced AI models have caused the administration to reevaluate this stance. Now, the White House is considering establishing an AI oversight board to review models before they are made publicly available.
THE IMPACT
The administration’s stance on AI has been evolving significantly over the past few months. Previously, the Trump administration prioritized innovation and competitiveness, oftentimes through a broad deregulatory posture. However, through Mythos, this stance is changing as the administration looks to take a more proactive approach to overseeing model releases and ensure that potentially market defining models have better controls placed on them.
Anthropic’s case underscores the importance of this shift where the administration was willing to ignore the AI developer and essentially blacklist them from federal agencies until Mythos was released. If Mythos is as operationally valuable as advertised, the Trump administration has strong incentives to de-escalate its conflict with Anthropic, most likely by finding some form of procedural off-ramp.
The Anthropic dispute likely marks an inflection point where the federal government is beginning to treat frontier AI models less like commercial software products and more like controlled strategic infrastructure.
Poland adopts new cryptocurrency regulations.
THE NEWS
On Friday, Poland passed a new law aimed at regulating cryptocurrencies. More specifically, the bill implements the European Union’s (EU) Markets in Crypto-Assets Regulation (MiCA). Notably, Poland was required to pass their local version of MiCA by the end of June, as the regional version is set to come into enforcement on July 1st, 2026.
For greater context, MiCA is an EU wide regulation designed to implement market rules for crypto-assets. Some of the key provisions in the law include regulations on:
- Issuing and trading crypto-assets
- Mandating stronger transparency and disclosure requirements
- The authorization and supervision of transactions
- Establishing a new legal framework to improve market integrity and financial stability
While Poland’s law is not different from the broader EU version, it does come at a time where the country is facing a potentially massive cryptocurrency scandal involving the Zondacrypto exchange and Russian foreign influence.
THE KNOWLEDGE
In March 2026, Polish prosecutors announced that they would be launching an investigation in to Zondacrypto over concerns regarding losses misleading clients and concerns over potential malign Russian influence.
While the cryptocurrency exchange’s boss, Przemyslaw Kral, has argued his innocence and assured that the currency is still solvent, thousands of Zondacrypto exchange users have reported being unable to withdraw their funds. Total losses are estimated to be more than 350 million Zlotys, or just over $95 million.
The Polish government has argued that the implications of this fraud probe are just financial losses, arguing that Russian money backed the firm and that there are concerns related to foreign political interference. In a government meeting earlier this month, Prime Minister Donald Tusk cited Polish security services when stating that:
“We were dealing with a company whose origins are particularly shady. It’s the Russian mafia and its money involved in organising the Zondacrypto exchange.”
Though the lawmakers have adopted the localized MiCA bill, its future is not certain. Previous attempts to pass the bill have been vetoed by President Karol Nawrocki and he could block the legislation again. If the nation does not pass the local version, Polish entities would lose the ability to provide crypto-asset services.
THE IMPACT
This investigation demonstrates how nations, especially within Europe, are beginning to pay closer attention to cryptocurrency markets. While MiCA was originally designed to standardize the EU’s crypto regulations and market rules, Poland’s case represents how exchanges could represent far more than just financial services but instead could be vectors for illicit finance, geopolitical pressures, and foreign influence.
For Poland specifically, this legislation and its rapidly approaching deadline is significant. If government investigators can link Zondacrypto to potential Russian actors, that could be a turning point and provide enough political pressure to pass the nation’s version of MiCA. If not, Poland could fall outside of the EU’s broader attempt to regulate cryptocurrencies, potentially creating great uncertainty for exchanges, investors, and other financial institutions.
This Week's Caveat Podcast: Scam ads, AI hallucinations, and legal implications.
Dave Bittner and Ben Yelin look at two stories. The first involves Santa Clara County suing Meta over alleged scam ads. In the lawsuit, the county claims that Meta allowed scam ads to filter through the system and deceive users to the company’s financial benefit. The second story involves a dismissed case where the plaintiffs attempted to sue Meta into removing a critical post. However, the case may end in sanctions for the lawyers whose arguments may have relied on fake AI sanctions.
OTHER NOTEWORTHY STORIES
CISA stores digital keys in open, plain text form.
What: The Cybersecurity and Infrastructure Security Agency (CISA) stored digital keys for its cloud storage accounts in open, plain-text form.
Why: On Monday, reports emerged that CISA was storing critical authentication data, including passwords, keys, tokens, and passwords, for its cloud storage accounts in plain-text in a .CSV file. It is unclear how long this information was unsecured, but the file was originally created in November 2025. In a statement, CISA emphasized:
“Currently, there is no indication that any sensitive data was compromised as a result of this incident. While we hold our team members to the highest standards of integrity and operational awareness, we are working to ensure additional safeguards are implemented to prevent future occurrences."
MAY 18, 2026 | Source: Gizmodo
US and China will start discussing AI safety.
What: Treasury Secretary Scott Bessent announced that the US and China will meet to discuss guardrails on AI.
Why: Last week, the Secretary Scott Bessent announced that the two sides will meet to outline AI guardrails, including establishing a protocol for keeping powerful models out of nonstate actors. However, it is unclear when these conversations will take place.
When announcing these future talks, Secretary Bessent stated:
“We’re going to set up a protocol in terms of, how do we go forward with best practices for AI to make sure nonstate actors don’t get a hold of these models.”
Secretary Bessent did note that the US is comfortable having these conversations as current US AI models are more advanced than Chinese alternatives.
MAY 14, 2026 | Source: New York Times
Anthropic to let partners share Mythos security findings.
What: Anthropic will allow Mythos users to share cyber threat information with others who could be exposed to similar vulnerabilities.
Why: On Monday, Anthropic announced it was revising its previous policy position to allow Mythos users to share critical vulnerability information with others that could likely be affected. With this policy change, an Anthropic spokesperson released a statement emphasizing:
“We fully support our partners sharing findings with each other and companies outside of Glasswing to triage vulnerabilities. While there was never a specific Glasswing NDA, confidentiality protections were something partners asked for at the outset and were built into agreements partners signed.”
Now, partners are allowed to share vulnerability information with security teams at other companies, industry bodies, government agencies, and open-source maintainers.
MAY 18, 2026 | Source: Reuters
UK moves to shield security researchers in cybercrime law overhaul.
What: The British government has announced that it plans to rewrite a key cybercrime law.
Why: Last week, the British government announced that it is planning to rewrite and update key cybercrime laws after warnings that the outdated laws were hindering security research and weakening cyber defenses. The proposed reforms would update the Computer Misuse Act 1990 as part of a larger national security package.
A spokesperson for the CyberUp Campaign commented on the announcement, emphasizing:
"Today marks a genuine turning point for cyber security in the UK. For years, the Computer Misuse Act has left legitimate cyber security professionals and researchers operating under unnecessary legal risk, while hostile actors move faster and with fewer constraints."
The government has not published any draft legislation.
MAY 13, 2026 | Source: The Record