In the autumn of 2026, the Federal Trade Commission turned its gaze toward two of artificial intelligence's most prominent architects — OpenAI and Anthropic — opening formal investigations into whether their most powerful systems place consumers in unreasonable danger. The move is less a sudden alarm than the culmination of a long-building tension between the speed of technological ambition and the slower, steadier demands of public accountability. Regulators are no longer content to watch from the margins; they are stepping into the room where the future is being built, asking to see the blue
FTC launches sweeping investigation into OpenAI, Anthropic over AI safety risks
The era of largely unregulated AI development is ending.
Mark
So the FTC is investigating OpenAI and Anthropic specifically. What makes these two companies the focus rather than, say, Google or Meta, which also have large language models?
Mimi
Both OpenAI and Anthropic are primarily focused on building and releasing general-purpose large language models—the kind of systems that can do many different tasks. That's different from Google or Meta, which have these models but use them mainly for their own products. OpenAI and Anthropic are in the business of selling access to their models directly to the public and to other companies. That makes them the most visible players in the space, and probably the ones regulators see as the most immediate consumer-facing risk.
Luke
But we should note the reporting doesn't actually say why the FTC picked these two over others. It's possible there are other investigations we don't know about yet, or it's possible the FTC has specific concerns about these companies' practices that aren't spelled out in the coverage.
Mark
What does "super intelligence" mean in this context? Is that a technical term or regulatory language?
Mimi
It's more of a descriptor that's entered common use. It refers to AI systems that are powerful and general enough that their behavior becomes hard to predict or control. The concern isn't that they're conscious or that they'll rebel—it's that they're complex enough that even their creators can't fully anticipate what they'll do in novel situations.
Luke
Right, and that's important: the term itself is somewhat contested. Some researchers use it to mean something very specific, others use it more loosely. The reporting doesn't define it, which means readers might have different ideas about what the FTC is actually worried about.
Mark
What would the FTC actually be looking for in an investigation like this?
Mimi
Typically, they'd want to see what safety testing the companies did before releasing systems, what risks they identified internally, whether they disclosed those risks to users, and whether they made claims about safety or capability that turned out to be false or misleading. They'd also look at whether the companies have adequate procedures for monitoring and responding to harms once systems are deployed.
Luke
The reporting doesn't specify what the FTC is actually demanding or what documents they're requesting. We know they've opened an investigation, but the scope and the specific allegations aren't detailed here. That's a real gap—it means we don't actually know yet whether this is about deceptive marketing, inadequate safety testing, or something else entirely.
Mark
Could this investigation actually change how these companies operate?
Mimi
Absolutely. If the FTC finds violations, it can impose penalties and require changes to practices. But even the investigation itself sends a signal—it tells the industry that regulators are watching and that safety claims need to be backed up by real procedures and documentation. Companies will likely become more cautious about what they claim their systems can do.
Luke
Though we should be careful not to overstate the impact. The FTC can enforce consumer protection law, but it can't write new regulations or set industry standards. That would require Congress or a dedicated regulatory agency. This investigation is important, but it's also limited in scope to what existing law allows.
The Pulse
- Federal regulators have formally escalated from observation to investigation, signaling that the era of AI development operating largely outside government scrutiny may be closing.
- The probe targets both OpenAI and Anthropic simultaneously — a deliberate choice that frames advanced AI risk not as one company's failing, but as a structural problem across the industry's leading edge.
- Safety researchers and consumer advocates have long warned that powerful models are being commercialized faster than their harms can be understood, and the FTC's action suggests those warnings have finally found institutional weight.
- Both companies, which have publicly championed AI safety as a core value, now face the harder test of demonstrating that commitment to regulators holding subpoena power and demanding internal documents.
- The investigation could set binding precedents — establishing what safety testing, risk disclosure, and consumer protections the entire AI sector must provide before deploying advanced systems at scale.
In the autumn of 2026, the Federal Trade Commission turned its gaze toward two of artificial intelligence's most prominent architects — OpenAI and Anthropic — opening formal investigations into whether their most powerful systems place consumers in unreasonable danger. The move is less a sudden alarm than the culmination of a long-building tension between the speed of technological ambition and the slower, steadier demands of public accountability. Regulators are no longer content to watch from the margins; they are stepping into the room where the future is being built, asking to see the blueprints.
The Federal Trade Commission has opened formal investigations into OpenAI and Anthropic, two of the most consequential developers of large language models, over concerns that their systems may pose unreasonable risks to consumers. The move represents a meaningful escalation — federal authorities shifting from passive observation of the AI industry toward active scrutiny with real enforcement tools behind it.
At the center of the inquiry are what regulators and researchers call 'super intelligence' models: AI systems powerful enough that their behavior and downstream effects become genuinely difficult to anticipate or contain. By targeting both companies at once, the FTC signals it views this as a sector-wide problem rather than an isolated failure at any single firm. OpenAI, the public face of generative AI through ChatGPT, and Anthropic, founded by former OpenAI researchers with safety as its stated north star, together represent the frontier the agency now wants to examine.
The investigation will likely require both companies to open their internal processes to scrutiny — safety testing protocols, risk assessments, communications about known harms, and the gap, if any, between what they say publicly and what they know privately. For Anthropic especially, which has built its identity around responsible development, the probe is a test of whether that identity holds under institutional pressure.
The consequences extend well beyond these two companies. If the FTC finds violations of consumer protection law, it can impose penalties or demand changes to how systems are deployed. More broadly, the investigation may effectively write the rules for the entire industry — establishing what documentation, testing, and disclosure regulators will require before advanced AI reaches the public. In the absence of comprehensive legislation, enforcement is becoming the rulemaking. For an industry accustomed to moving fast, that is a new and unfamiliar kind of gravity.
The Federal Trade Commission has opened a formal investigation into OpenAI and Anthropic, two of the most prominent developers of large language models, focusing on whether their systems pose unreasonable risks to consumers. The move marks a significant escalation in regulatory scrutiny of the artificial intelligence industry and signals that federal authorities are moving beyond observation into active enforcement.
The investigation centers on safety concerns tied to what regulators and industry observers call "super intelligence" models—AI systems of sufficient power and sophistication that their behavior and impact become difficult to predict or contain. The FTC's decision to probe both companies simultaneously suggests the agency views this not as an isolated problem at one firm but as a systemic issue across the sector's leading players. OpenAI, which developed ChatGPT and has become the public face of generative AI, and Anthropic, a newer competitor founded by former OpenAI researchers, represent the cutting edge of the technology the FTC now wants to examine.
The timing of the investigation reflects mounting pressure from lawmakers, safety researchers, and consumer advocates who have raised alarms about the deployment of powerful AI systems without adequate safeguards or transparency. These critics argue that companies have moved quickly to release and commercialize advanced models while the potential harms—from misinformation and privacy violations to more speculative but serious risks—remain poorly understood. The FTC's action suggests the agency agrees that waiting for harm to materialize before acting is no longer acceptable.
What the investigation will actually examine remains somewhat opaque at this stage. The FTC typically uses such probes to gather information about a company's practices, safety testing protocols, and risk assessment procedures. The agency may seek to understand how these companies evaluate potential harms before deployment, what safeguards they have implemented, and whether they have adequately disclosed risks to users. The investigation could also examine whether companies have made misleading claims about their systems' capabilities or safety.
The probe carries real consequences for how the AI industry operates going forward. If the FTC finds violations of consumer protection law, it can impose penalties, demand changes to business practices, or seek injunctions against certain activities. More broadly, a high-profile investigation into major AI developers could establish expectations for the entire sector—essentially setting a floor for what regulators will demand in terms of safety documentation, testing, and disclosure before companies can deploy new systems at scale.
For OpenAI and Anthropic, the investigation represents a new kind of pressure. Both companies have positioned themselves as taking AI safety seriously, with Anthropic in particular founded on the premise that safety should be central to AI development. Now they will have to demonstrate that commitment to skeptical regulators armed with subpoena power. The companies will likely need to produce internal documents, safety assessments, and communications about known risks—the kind of material that could either vindicate their safety practices or expose gaps between their public statements and internal concerns.
The investigation also reflects a broader shift in how government approaches technology regulation. Rather than waiting for comprehensive legislation to pass, the FTC is using its existing authority under consumer protection law to intervene in real time. This approach allows faster action but also means the rules are being written through investigation and enforcement rather than through transparent rulemaking. For the AI industry, it creates uncertainty about what exactly will be required, but it also signals that the era of largely unregulated AI development is ending.
Notable Quotes
Both companies have positioned themselves as taking AI safety seriously, with Anthropic in particular founded on the premise that safety should be central to AI development.— Reporting on company positioning
Want the full story?
Read the original at Google News ↗