The Trump administration has filed an appeal against a federal court ruling that temporarily blocked the Pentagon from designating Anthropic as a supply chain risk. The move escalates a legal dispute that underscores growing tensions between AI developers and government agencies over safety standards and operational control.

The appeal follows a decision by U.S. District Judge Rita Lin, who sided with Anthropic and halted both the supply chain risk designation and a broader directive requiring federal agencies to sever ties with the company. The directive, if enforced, would prevent Anthropic from securing government contracts and restrict companies working with the military from partnering with the firm.

Judge Lin delayed the implementation of her ruling by one week, allowing the administration time to seek relief through the appeals process. The government’s response was widely anticipated given the implications of the decision for federal procurement and national security policy.

Anthropic initiated legal action after the Pentagon labeled the company a supply chain risk, reportedly following disagreements over AI safety conditions. The company has maintained that its technology should not be used in fully autonomous lethal weapons or for large-scale domestic surveillance.

In its complaint, Anthropic argued that the Pentagon’s actions were retaliatory and violated its constitutional rights. The company claimed it was penalized for expressing a “protected viewpoint” on the ethical use of artificial intelligence.

Broader Implications for AI Regulation

Judge Lin indicated that Anthropic is likely to succeed in its claims, citing concerns that due process requirements were not properly followed by the Department of Defense. The ruling has drawn attention across the technology sector, where companies are increasingly navigating complex relationships with government agencies.

Anthropic has previously partnered with the Pentagon, reflecting a trend in which AI firms collaborate with government agencies while attempting to maintain internal safeguards on how their technologies are deployed.

The outcome of the appeal could set a precedent for how far governments can go in restricting private AI companies based on policy disagreements. It may also influence how future contracts between AI developers and defense agencies are structured, particularly regarding usage limitations and compliance requirements.

AI & Machine Learning, News, Regulation & Policy

OpenAI is introducing a new pricing structure for Codex, its AI coding agent, aimed at making adoption more flexible for enterprise teams. Starting today, organizations using ChatGPT Business and Enterprise can add Codex-only seats with pay-as-you-go pricing, removing the need for fixed per-seat fees.

The update allows teams to access Codex without upfront commitments, enabling smaller groups to run targeted pilots before scaling usage across the organization. Instead of subscription-based pricing, usage is billed based on token consumption, offering more transparency into how activity translates into cost.

Codex-only seats also remove rate limits, allowing unrestricted usage based on demand. This model is designed to give teams greater control over budgeting, particularly for engineering workflows that may vary in intensity over time.

At the same time, OpenAI is adjusting its broader pricing strategy. The annual cost of ChatGPT Business has been reduced from $25 to $20 per seat, making standard access more affordable for organizations that want bundled features, including limited Codex usage.

To further encourage adoption, OpenAI is offering promotional credits for new Codex users. Eligible ChatGPT Business workspaces can receive $100 in credits per new Codex-only user, up to $500 per team, as part of a limited-time incentive.

Rapid Growth in Developer Usage

The pricing changes come as Codex adoption accelerates across enterprise environments. OpenAI reports that more than 2 million developers are now using Codex weekly, while overall ChatGPT business users exceed 9 million. Within Business and Enterprise tiers, Codex usage has grown sixfold since the beginning of the year.

Companies including Notion, Ramp, Braintrust, and Wasmer are using Codex to streamline software development workflows. Common use cases include automating repetitive coding tasks, generating production-ready code, and improving collaboration across engineering teams.

The company is also expanding Codex’s integration capabilities. New features such as plugins and automations allow teams to connect Codex with internal tools and external systems, enabling more complex and coordinated workflows. Dedicated applications for macOS and Windows further support adoption by embedding the tool directly into developer environments.

The shift toward usage-based pricing reflects a broader trend in enterprise AI, where companies are moving away from fixed licensing models toward consumption-based billing. This approach aligns costs more closely with value delivered, particularly for tools that are used intermittently or scale dynamically across teams.

AI & Machine Learning, News

OpenAI has acquired Technology Business Programming Network (TBPN), a fast-growing tech media platform, as part of a broader effort to reshape how it communicates about artificial intelligence. The company said the acquisition is aimed at strengthening engagement with developers, businesses, and the wider public as AI adoption accelerates globally.

TBPN has built a strong following through its daily live programming focused on technology, startups, and AI developments. The platform is known for convening industry voices and providing real-time commentary on major announcements across the tech ecosystem.

OpenAI plans to integrate TBPN into its Strategy organization, where the team will report to Chris Lehane. The move reflects a shift toward more direct and continuous communication channels, rather than relying solely on traditional corporate messaging.

According to OpenAI leadership, the scale and pace of AI development require new approaches to public engagement. The company is positioning TBPN as a platform to facilitate ongoing discussions about the impact of AI, particularly among builders and users of the technology.

A key condition of the acquisition is maintaining TBPN’s editorial independence. The platform will continue to manage its programming, guest selection, and content decisions independently, a structure OpenAI said is essential to preserving credibility and trust within the tech community.

Strategic Role in the AI Ecosystem

Beyond content, OpenAI expects TBPN to contribute to its broader communications and marketing efforts. The media team’s experience in digital engagement and audience development is expected to support how OpenAI presents its products and research to a global audience.

TBPN’s founders and leadership team, including Jordi Hays, John Coogan, and Dylan Abruscato, will join OpenAI as part of the transition. The company noted that TBPN’s ability to capture industry sentiment and translate complex developments into accessible discussions aligns with its long-term goals.

The acquisition also reflects increasing competition among AI companies not only in technology development but in shaping public understanding of the field. As AI systems become more embedded in daily life, communication strategies are emerging as a key differentiator alongside model performance and infrastructure.

TBPN stated that its move from independent commentary to direct involvement in AI distribution and communication represents an opportunity to have a more tangible impact on how the technology is understood globally.

AI & Machine Learning, News, Startups & Investment

Anthropic inadvertently caused widespread disruption on GitHub after issuing a copyright takedown request aimed at removing leaked source code from its Claude Code application. The notice, filed under U.S. digital copyright law, resulted in approximately 8,100 repositories being taken offline, including many unrelated or legitimate projects.

The issue originated when a recent release of Claude Code, Anthropic’s command-line developer tool, unintentionally exposed portions of its underlying source code. The leak quickly circulated among developers and AI enthusiasts, with copies appearing across multiple GitHub repositories.

In response, Anthropic filed a takedown request to remove the leaked code. However, due to the structure of GitHub’s repository forking system, the request extended beyond the intended targets. It affected not only repositories hosting the leaked material but also legitimate forks connected to Anthropic’s own public codebase.

The scale of the takedown prompted immediate backlash from developers who found their repositories inaccessible despite containing no sensitive or infringing content.

Company Reverses Action Amid IPO Preparations

Anthropic later acknowledged that the scope of the takedown was broader than intended. According to the company, the repository identified in the request was part of a larger fork network, which led to the unintended removal of thousands of related repositories.

The company has since retracted most of the takedown notices, limiting enforcement to a single repository and a smaller set of confirmed forks containing the leaked code. GitHub has restored access to the majority of affected repositories following the correction.

See more

This was not intentional, we’ve been working with GitHub to fix it. Should be better now.

— Boris Cherny (@bcherny) April 1, 2026

The incident highlights operational challenges associated with managing proprietary AI systems in open development environments. As companies increasingly release tools and APIs to developers, the boundary between public and private codebases can become more complex to manage.

The timing of the error is notable as Anthropic is reportedly preparing for a potential initial public offering as soon as October this year. Such incidents may draw increased scrutiny from investors, particularly around internal controls, compliance processes, and intellectual property management.

AI & Machine Learning, News

SpaceX has taken a significant step toward going public, reportedly submitting a confidential filing for a U.S. initial public offering. The move positions Elon Musk’s rocket and satellite company at the forefront of a new wave of high-profile listings expected to reshape public markets in 2026.

If completed, the offering could become the largest IPO in history. Analysts estimate that a raise exceeding $25.6 billion would surpass Saudi Aramco’s record-setting 2019 debut. The potential listing comes as SpaceX continues to expand its role beyond aerospace, particularly following its acquisition of Musk’s artificial intelligence startup xAI earlier this year.

The integration of xAI signals a broader strategy to combine space infrastructure with AI capabilities, including data processing and satellite-enabled services. This convergence reflects growing investor interest in companies that operate across both physical and digital infrastructure layers.

Wall Street expects 2026 to mark a strong recovery in IPO activity after several subdued years. Goldman Sachs has projected that U.S. IPO proceeds could reach as much as $160 billion, driven by pent-up demand and a backlog of large private companies preparing to go public.

However, market conditions remain uncertain. Geopolitical tensions and ongoing volatility in global equities could influence timing and valuations for major listings.

AI Companies Prepare for Public Market Debuts

Alongside SpaceX, leading artificial intelligence firms are also laying the groundwork for potential IPOs. OpenAI is reportedly exploring a public listing that could value the company at up to $1 trillion, reflecting its rapid growth and central role in the generative AI market. While the company has previously indicated that an IPO is not imminent, preparations suggest a longer-term path toward public markets.

Anthropic, the developer of the Claude AI model, is also preparing for a potential listing in October. The company has engaged legal advisors as part of early-stage IPO planning, with reports indicating a possible debut as soon as 2026.

These developments highlight a broader shift in the technology sector, where AI companies are transitioning from research-focused organizations to large-scale commercial platforms. With increasing revenue, enterprise adoption, and infrastructure investments, firms like OpenAI and Anthropic are positioning themselves to meet public market expectations.

The convergence of AI and capital markets reflects rising demand for exposure to next-generation technologies. As companies scale their models and expand into enterprise applications, IPOs are emerging as a key mechanism to fund continued growth.

If market conditions stabilize, the simultaneous entry of SpaceX, OpenAI, and Anthropic could define the next phase of the tech IPO cycle, with AI-driven businesses at its core.

AI & Machine Learning, News, Startups & Investment