Over less than two weeks, OpenAI has simultaneously shipped one of its most compelling education features to date and found itself confronting legal, political, and financial pressures on multiple fronts. The company’s new interactive math and science tools for ChatGPT land at a moment when its relationships with governments, users, and even its own staff are under unusual strain.
Inside OpenAI’s new interactive math and science tools
The latest ChatGPT update introduces a set of interactive visual modules designed to help users understand core mathematical and scientific concepts by manipulating variables in real time. OpenAI says the experience now covers more than 70 topics, spanning high school and introductory college material—from the Pythagorean theorem and Ohm’s law to compound interest, exponential decay, kinetic energy, and gas laws.
When a user asks ChatGPT to explain one of these concepts (for example, “help me understand the Pythagorean theorem”), the chatbot responds with a written explanation accompanied by a dynamic panel. Sliders let users adjust inputs such as the lengths of sides a and b in the equation a² + b² = c²; as values change, the computed hypotenuse and geometric visualization update instantly. Similar treatments appear for voltage and resistance in Ohm’s law, pressure and temperature in the ideal gas equation, and geometric quantities like cone and cylinder volume or surface area of a sphere.
OpenAI is explicitly leaning on established pedagogy: students often grasp formulas more effectively when they can see how changing one variable transforms the entire system. The company points to research indicating that visual, interaction-based learning can yield stronger conceptual understanding than traditional instruction for many learners, and it cites a recent Gallup survey finding that more than half of U.S. adults say they struggle with math.
In early testing, according to OpenAI, students reported that the interactive modules helped them see relationships between variables more clearly, while some parents described using the tools alongside their children. High school math teachers quoted in the company’s announcement emphasized the focus on conceptual understanding and the potential to help students explore abstract ideas independently.
Crucially, the feature is not paywalled: it is available to all logged-in ChatGPT users worldwide across every plan, including free. That broad reach matters because OpenAI says 140 million people already use ChatGPT each week for math and science learning—a scale that makes any change to its learning experience consequential for educators and policymakers alike.
How ChatGPT’s education push fits into OpenAI’s strategy
The interactive tools extend a line of education-focused capabilities OpenAI has been building into ChatGPT. The service already offers a “study mode” for step-by-step problem solving and a quizzes feature aimed at exam preparation. The new modules add a layer of visualization and direct manipulation that moves the product further away from static Q&A and closer to an exploratory learning environment.
Strategically, education remains OpenAI’s cleanest and least controversial use case. It is one of the few domains where AI’s value proposition—helping people understand complex material—is broadly intuitive and less entangled with concerns about surveillance, disinformation, or labor displacement. The company emphasizes that it intends to expand interactive learning to additional subjects and to publish research through its NextGenAI initiative and OpenAI Learning Lab on how AI changes learning outcomes over time. OpenAI itself notes that the research landscape on AI in education is still emerging, and so far it characterizes its own findings from study mode as “promising early signals” rather than definitive results.
Within the competitive landscape, these real-time visualizations give ChatGPT a differentiator. Rivals such as Google’s Gemini, Anthropic’s Claude, and xAI’s Grok are all investing in education applications, but according to OpenAI’s description, none currently offer the same style of tightly integrated, interactive formula manipulation inside a conversational interface.
For educators and policymakers, this positioning matters. If ChatGPT becomes a default companion for math and science learning at scale, questions around curriculum alignment, assessment integrity, and equity of access will intensify. OpenAI’s decision to make the tools available to free users lowers an immediate access barrier, but it also ensures that any design decision—such as how concepts are explained or visualized—can influence how millions of students worldwide encounter foundational STEM ideas.
The Tumbler Ridge lawsuit and questions of AI duty of care
The timing of the education launch is notable because it came a day after OpenAI was hit with what may be its most serious legal challenge to date. A lawsuit filed in British Columbia Supreme Court by the mother of 12-year-old mass shooting victim Maya Gebala alleges that OpenAI had “specific knowledge of the shooter’s long-range planning of a mass casualty event” gleaned from ChatGPT interactions and “took no steps to act upon this knowledge.”
The February 10 attack in Tumbler Ridge, British Columbia, left eight people dead, including the 18-year-old shooter, and Gebala with a catastrophic traumatic brain injury and permanent cognitive and physical disabilities, according to the claim. The filing portrays ChatGPT as having functioned for the shooter as “counsellor, pseudo-therapist, trusted confidante, friend, and ally” and asserts that the system was “intentionally designed to foster psychological dependency” on the service.
The lawsuit also argues that, although OpenAI formally requires minors to obtain parental consent, it “took no steps to implement age verification or consent procedures” in this case. OpenAI has acknowledged separately that it suspended the shooter’s account months before the attack but did not notify Canadian law enforcement—a decision that triggered political backlash. British Columbia Premier David Eby, after speaking with OpenAI CEO Sam Altman, said Altman had agreed to apologize to the community and collaborate on AI regulation recommendations.
None of the allegations have been tested in court, and OpenAI has not publicly commented on the substance of the lawsuit. Nonetheless, the case foregrounds a difficult question for AI providers: once an internal system identifies a user as sufficiently dangerous to warrant an account suspension, what obligations—legal or ethical—follow? The answer will shape expectations around AI platforms’ responsibility to monitor and report user behavior, particularly when minors are involved.
Inside the Pentagon deal, internal dissent, and competitive fallout
The legal scrutiny arrives as OpenAI navigates an internal and external backlash over its decision to provide the U.S. Department of Defense with access to its AI models on classified networks. On February 28, Altman announced an agreement allowing the Pentagon to deploy OpenAI technology inside secure government computing environments.
The deal contrasted sharply with rival Anthropic’s stance. Anthropic CEO Dario Amodei had publicly declined similar terms without guarantees against uses involving autonomous weapons or mass domestic surveillance. The Pentagon responded by labeling Anthropic a “supply-chain risk”—a designation typically applied to foreign adversaries—and Defense Secretary Pete Hegseth barred military contractors from commercial activity with the company.
Inside OpenAI, the reaction was swift. Caitlin Kalinowski, head of robotics hardware and a former Meta executive, resigned in protest, arguing in a public statement that AI’s role in national security needed firmer lines around surveillance without judicial oversight and lethal autonomy without human authorization than the deal reflected. Other researchers expressed misgivings on social media, and at least one employee quoted in press reports said many staff “really respect” Anthropic for walking away.
Externally, the reputational cost was immediate. ChatGPT app uninstalls surged almost 300% on the day the Pentagon partnership was announced. Anthropic’s Claude app rose to the top of the U.S. Apple App Store’s free apps chart and was still there days later. Protesters gathered outside OpenAI’s San Francisco headquarters, rallying around a “QuitGPT” campaign urging users to delete the app.
In an unusual development, more than 30 employees from OpenAI and Google DeepMind—including DeepMind chief scientist Jeff Dean—filed an amicus brief supporting Anthropic’s lawsuit challenging the Pentagon’s treatment. Signed in a personal capacity, the brief argued that the Defense Department’s actions, if allowed to stand, could harm U.S. industrial and scientific competitiveness in AI. The spectacle of OpenAI researchers backing a competitor’s case against the same government department their own employer had just partnered with underlines the depth of the industry’s divisions over military AI work.
Altman has acknowledged missteps. In an internal memo later made public, he described the Pentagon agreement as “definitely rushed” and said it “just looked opportunistic and sloppy.” OpenAI updated the contract to add explicit prohibitions on mass domestic surveillance and on using its technology with commercially acquired data. Altman also publicly criticized the Pentagon’s blacklisting of Anthropic, saying that enforcing the supply-chain risk designation would be harmful to the broader industry and the country.
Anthropic, for its part, has told the court that the Pentagon’s action threatens up to $5 billion in business—roughly equal to its total revenue since commercializing its AI systems in 2023—and is seeking a temporary order allowing it to continue working with defense contractors while the case proceeds.
Cash burn, ads, and infrastructure: why user trust now has a price tag
Beneath the political and legal controversies, OpenAI faces a straightforward financial challenge. The company is expected to burn approximately $15 billion in cash this year, up from $9 billion in 2025. With around 910 million weekly users, and roughly 95% of them on free tiers, subscription revenue alone is insufficient to cover the gap.
To close it, OpenAI is building an advertising business inside ChatGPT, working with partners such as Criteo and, reportedly, The Trade Desk while simultaneously standing up its own ad-tech stack. Hiring plans underscore the ambition: roles span monetization infrastructure engineering, ads product design, ad revenue accounting, and trust and safety focused specifically on advertising, with compensation bands reaching into the high six figures. This is the profile of a company that intends to own, not outsource, its ad infrastructure.
However, introducing ads into ChatGPT adds another layer to an already complex trust equation. The Pentagon deal showed that user loyalty can be fragile; the surge in uninstalls suggests that many people are willing to move quickly when they perceive misalignment with their values. Adding commercial messaging into a product that is simultaneously facing questions about its handling of sensitive data in the Tumbler Ridge case creates a compounded risk: any perception that conversation outputs are shaped by advertisers or other external pressures could erode confidence in the system’s neutrality.
On the infrastructure side, OpenAI’s decision with Oracle to abandon plans for a major AI data center expansion in Abilene, Texas, after disagreements over financing and evolving technical requirements, left a vacuum that competitors moved quickly to explore. Meta and Nvidia reportedly showed interest in the site—illustrating how, in the current AI build-out, any lapse in execution can quickly advantage rivals with similar capital-intensive needs.
In that context, retaining and deepening user engagement is not just a product goal; it is a financial necessity. Every decision that affects how people feel about using ChatGPT—whether a Pentagon contract, a lawsuit over content moderation, or the presence of ads—feeds back into OpenAI’s ability to sustain its current burn rate and infrastructure buildout.
Education as OpenAI’s most persuasive narrative—for now
Against this backdrop, the new interactive learning tools function as more than a product update. They are a statement about what OpenAI wants to be seen as building. Education is where ChatGPT most clearly augments human capability: helping a student visualize a geometry problem, supporting a parent who has forgotten the details of algebra, or giving adults a way to revisit concepts they struggled with in school.
OpenAI underscores that it will continue partnering with educators and researchers through its NextGenAI initiative and Learning Lab, and that it plans to publish findings as it expands to more subjects. At the same time, the company concedes that the broader research picture on AI’s impact on learning outcomes is still forming. For policymakers and school systems considering formal adoption or guidance, that uncertainty means the tools’ promise must be weighed alongside emerging evidence on long-term effects and potential dependencies.
The juxtaposition is stark: on one screen, sliders move and graphs shift, helping a student finally see why the Pythagorean theorem works. Off-screen, the same company is arguing about military partnerships, facing a wrongful death-related lawsuit, dealing with public protest, and funding all of it with a projected $15 billion in annual cash burn. For many of the 140 million weekly math and science learners, those contexts will remain invisible; they will simply experience a system that, at least in that moment, helps something “click.”
For industry professionals, regulators, and educators, the challenge is different: to evaluate how much weight to give those genuine learning gains when set against the unresolved questions of oversight, accountability, and business model pressure that now surround one of the world’s most influential AI platforms.
Hi, I’m Cary Huang — a tech enthusiast based in Canada. I’ve spent years working with complex production systems and open-source software. Through TechBuddies.io, my team and I share practical engineering insights, curate relevant tech news, and recommend useful tools and products to help developers learn and work more effectively.
