8-minute read | 1,650 words
What to know this week
EU fines AliExpress.
The European Commission has fined AliExpress 550 million euros for risk management failures.
OpenAI backs new AI safety bill.
OpenAI has backed a Massachusetts AI safety bill that would add mandatory audits for AI developers.
This week's full stories
EU issues largest DSA fine against AliExpress.
THE NEWS
On Monday, the European Commission announced that it fined AliExpress, the online marketplace 550 million euros. When issuing the fine, the European Commission emphasized that the company had failed to assess and mitigate the risks created by the illegal, unsafe, and counterfeit products being sold on its site.
More specifically, the Commission stated that AliExpress failed to properly evaluate if it had enough resources in place to monitor for illegal products and overstated the effectiveness of its detection and removal systems. Additionally, the Commission alleged that the company did not adequately consider how its recommendation and advertising tools would increase the visibility of illegal products.
Henna Virkkunen, the Commission’s Vice President for technology sovereignty, security, and democracy commented on the fine, stating:
“The spread of counterfeit clothing, unsafe toys, dangerous cosmetics and other illegal and harmful products is not an unavoidable cost of shopping online. Scale is not an excuse; risks must be identified and addressed systematically to ensure consumers can safely shop online.”
The penalty is the largest fine issued under the EU’s Digital Services Act.
THE KNOWLEDGE
Alongside the European Union’s (EU) fine against AliExpress, other major Chinese e-commerce platforms have also been targeted. In May 2026, the Commission fined Temu 200 million euros for breaching the DSA. In this incident, the Commission took issue with Temu’s 2024 risk assessment. The Commission argued that Temu’s risk assessment fell short of the DSA’s standards as it:
- Based its information about risks concerning eCommerce as a whole, rather than on evidence related to Temu’s own service.
- Underestimated how often EU consumers encounter illegal items.
- Did not properly assess how its service could amplify dissemination risks of legal products.
With these findings and subsequent fine, the EU gave Temu until August 28, 2026 to submit a new action plan to rectify the Commission’s concerns.
Another major Chinese eCommerce platform, Shein, is facing similar scrutiny. In February 2026, the Commission announced that it was launching an investigation into Shein for DSA violations. For this investigation, the Commission stated that it was focusing on:
- The mechanisms Shein uses to limit the sale of illegal products in the EU.
- The risks related to any of the platform’s addictive design features.
- The transparency and Shein’s recommender systems.
The Commission also emphasized that this investigation was launched after completing a preliminary analysis of the risk assessment reports provided by Shein and from information shared by third parties.
In conjunction with these investigations and fines, the EU is also looking to limit the region’s influx of “low-value” imports. Beginning in July 2026, the EU announced a new levy against small parcels entering the bloc. The new law levies a three euro tax on all parcels worth less than 150 euros. The new tax was specifically established to curb the growing market dominance of the Chinese e-commerce markets previously discussed.
THE IMPACT
With each of these investigations, the Commission has consistently framed these actions as efforts to better protect consumers from illegal, counterfeit, and unsafe products. However, the continued focus on Chinese eCommerce platforms also reflects another policy agenda for the EU’s economic security. As Chinese marketplaces have rapidly expanded across Europe, EU policymakers are now looking to ensure that these companies are operating under the same standards that domestic competitors follow.
Taken together, these investigations and new levies suggest that the EU is using digital regulation not only to improve safety for consumers but also to strengthen its own economic resilience against foreign market influence. As geopolitical competition increasingly intersects with commerce, policymakers are likely to continue using these enforcement tools to reduce dependencies and minimize foreign economic influence.
OpenAI backs Massachusetts AI safety efforts.
THE NEWS
On Tuesday, OpenAI expressed its support for a Massachusetts Senate bill that aims to mandate new safety requirements for AI development.
In a statement, OpenAI spokesperson Liz Bourgeois said the company supports state efforts to establish AI safeguards in the absence of comprehensive federal action. She reiterated OpenAI’s support for its overall strategy dubbed “reverse federalism.”
This concept involves working with states to adopt similar AI regulations that can lead to establishing a de facto national framework without requiring Congress to pass a single federal law. Bourgeois added:
“In the absence of federal action, we support state efforts like this one in Massachusetts - alongside legislation already in place in California, New York, and Illinois - helping to create a de facto nationwide approach.”
The support comes as Massachusetts lawmakers in the state’s House are considering another AI regulation. Anthropic has backed a separate, more restrictive proposal that includes requirements for independent evaluations of potential catastrophic risks and additional enforcement mechanisms.
THE KNOWLEDGE
These two bills highlight a growing divide in how major AI developers are attempting to shape state-level AI policies
Anthropic has advocated for states to adopt increasingly ambitious AI safety requirements. In a recent interview, the company’s head of US state and local government relations, Cesar Fernandez, noted that Anthropic wants legislation that “meaningfully raises the bar on safety” for most AI systems.
Meanwhile, OpenAI has promoted its strategy of “reverse federalism,” which seeks to encourage states to unite around a common set of AI safeguards. By creating this state-aligned approach, OpenAI hopes to avoid creating a regulatory patchwork that could result from fifty different state approaches.
Both strategies involve trade-offs.
Anthropic’s approach could encourage states to adopt stronger safeguards and establish more ambitious standards for managing the risks posed by AI systems. However, more restrictive laws often face greater political and legal challenges. Furthermore, if states adopt significantly different AI regulations, companies would likely face a more complicated and burdensome compliance environment.
OpenAI's approach could make it easier for states to establish a consistent national baseline, likely reducing compliance costs and promoting greater innovation. However, a common framework also limits states abilities to adopt stronger requirements or experiment with more ambitious legislation, which could improve safety.
THE IMPACT
While it is unclear which law the Massachusetts state legislature will pass, if any, both laws are representative of a growing policy debate. Should states converge around a common regulatory framework, or should individual states continue pushing AI policy boundaries individually?
Since the federal government has yet to establish a comprehensive federal framework for AI safety, state governments have increasingly become an important testing ground for new regulations. That approach could allow policymakers to experiment with different strategies and potentially create stronger safeguards. However, it also creates the possibility of a fragmented regulatory environment in which AI developers and deployers must comply with radically different requirements across the nation.
As major AI developers increasingly engage with state policymakers, the competing strategies of OpenAI and Anthropic will be important. The outcome could help determine whether the US develops a relatively consistent national AI safety baseline through state convergence, or a more fragmented, individualized system.
This Week's Caveat Podcast: AI’s political and copyright reckoning.
Dave Bittner and Ben Yelin discuss two key stories. The first involves how political campaigns are having to deal with AI chatbots producing unflattering or incomplete information when voters ask candidate questions. The second story involves how current copyright laws are not able to contend with how Generative AI ingests content and the need for new legal standards.
OTHER NOTEWORTHY STORIES
Head of US AI safety resigns.
What: Director Chris Fall from the US Center for AI Standards and Innovation has resigned after three months.
Why: On Monday, Director Chris Fall resigned from his position after serving for three months. Arvind Raman, from the Commerce Department, will fill the role temporarily. The Commerce Department did not provide a reason for Fall’s departure.
This institute is responsible for leading AI labs to test their unreleased models for vulnerabilities. For example, the agency focuses on calculating the “demonstrable risks” posed by advanced AI models. The agency works with leading AI labs such as Anthropic, OpenAI, and Google’s DeepMind.
JULY 20, 2026 | Source: Reuters
France approves social media ban
What: France passed a new law banning users under the age of fifteen from accessing social media platforms.
Why: On Wednesday, both chambers of France’s Parliament voted in favor of this social media ban. Additionally, the ban also bars the use of mobile phones in high schools.
France’s President, Emmanuel Macron, has been an advocate for passing this ban before his tenure ends next year. When speaking on the measure, Macron stated:
“France is leading the way in Europe in protecting our children and our teenagers. We will keep on going.”
This ban does not cover online encyclopedias or any educational directories.
Ines Legendre, a legal adviser for online protection group e-Enfance, did note that there are hurdles that still need to be addressed. Legendre stated:
“We’ll also have to address the issue of existing accounts for those under 15. How do we identify them? How do we suspend them? And then there’s the question of age verification for all new accounts that will come into effect.”
JULY 22, 2026 | Source: NPR
Meta faces Tennessee trial over addictive design features.
What: A lawsuit between Meta and Tennessee started over the state’s claims that Instagram’s design significantly contributed to the youth mental-health crisis.
Why: On Monday, Tennessee’s lawsuit against Meta began as the state alleges that Instagram’s design is to blame for a youth mental-health crisis. In the suit, Tennessee alleges that the company violated the state’s consumer protection law by knowingly creating a product that drove compulsive use behaviors.
The state also alleges that CEO Mark Zuckerberg was repeatedly warned about the negative impacts on teens but declined to fund those efforts and made misleading public statements.
JULY 20, 2026 | Source: Reuters