AI × Law

Coinbase, Shopify and Ramp have built internal coding AI agents for employees instead of relying only on tools from OpenAI or Anthropic. The article says the move is driven by rising AI spend and shows that some companies now want to control the orchestration layer between developers and foundation models.

A German court ruled that using a copyrighted photograph to create a sufficiently transformed AI-generated image did not amount to copyright infringement. The decision is notable for litigators because it touches the boundary between source images and AI outputs in copyright disputes.

A German court ruled that AI music company Suno infringed copyright by using works represented by collecting society GEMA without permission. The ruling puts a spotlight on whether training data itself can create infringement risk, not just the songs AI tools generate, and could pressure AI music providers to clarify licensing and permissions.

A U.S. District Court in Minnesota denied xAI’s request for a temporary restraining order against the state’s new nudification law, clearing the way for the ban to take effect Aug. 1. The case is notable because it pits an AI company against a state law targeting nonconsensual image-manipulation tools and raises First Amendment issues around AI-generated content.

Experts say trademark clients are increasingly demanding proof of value and pushing back on traditional billing models because of AI. The shift matters to legal professionals because it suggests AI is starting to change how IP work is priced and sold, not just how it is done.

Fifteen state attorneys general sent OpenAI a letter demanding preservation of all potentially relevant materials tied to the July security incident. The article says the letter also seeks a halt to certain internal AI cyber-testing and warns of possible spoliation sanctions if litigation follows, underscoring growing legal exposure around agentic AI security failures.

Apple and OpenAI are trading fresh court filings over Apple’s lawsuit accusing OpenAI of using former Apple employees to steal trade secrets for a secretive consumer device project. OpenAI is asking the court to permanently dismiss the case, while Apple says the alleged misuse is ongoing and seeks an injunction; a hearing is set for Oct. 1. The dispute is notable because the companies are also partners, with ChatGPT integrated into Apple products since 2024.

Courts in Australia, the UK, and the US are developing distinct approaches to address how using generative AI impacts legal professional privilege and client confidentiality. Key concerns include inadvertent waiver when uploading privileged material to public AI platforms and the potential for unverified AI outputs to cause disclosure issues.

K&L Gates lawyers are pro bono developing and refining an AI-powered chatbot, LIA (Legal Information Assistant), for Legal Aid of North Carolina. The tool, live on Legal Aid's website, helps address overwhelming demand for legal services by providing users with legal information and guidance, leveraging AI models like OpenAI's ChatGPT and Anthropic's Claude.

Australian legal tech company Contractam has published a 'Responsible AI' page, outlining how its AI analyzes contracts for legal compliance, fairness, and clarity. The document specifies the AI's reliance on live legal data, its self-correction mechanisms, and critical limitations such as its inability to interpret unstated facts or guarantee legal outcomes.

Legal tech startup BrentWorks introduced CiteSentinel, a platform designed to identify and prevent AI-generated hallucinations in legal citations. The tool scans legal documents to flag fabricated or erroneous case law, statutes, and authorities, helping lawyers avoid sanctions and embarrassment.

AI-powered legal technology startup Norm AI has successfully closed a $120 million Series C funding round led by Khosla Ventures, elevating its valuation to $1.2 billion. This significant investment positions Norm AI as a key player in the legal tech market.

Legal AI startup Harvey AI is reportedly seeking $500 million in new funding, pushing its valuation to $15.5 billion, a 41% increase in six months. This comes as the company's annualized revenue soared to $350 million, signaling strong adoption of legal AI among firms and corporate legal departments. Harvey also plans to develop its own foundation models, potentially impacting its relationships with current third-party AI providers.

A new survey details eDiscovery pricing trends, showing stabilization in traditional services like forensic collection while GenAI-assisted review introduces diverse, still-developing commercial models. Litigators and legal professionals can use this data for budgeting, vendor selection, and risk management, as AI's impact on costs and efficiency becomes clearer yet unevenly adopted across the industry.

A South Korean court has ordered a deepfake creator to pay damages to a victim whose likeness was used without consent. This ruling signals a growing legal precedent for accountability in the creation and dissemination of AI-generated content, impacting future litigation around digital likeness rights.

This checklist details legal, technical, and organizational controls for AI-powered biometric systems, focusing on compliance with the EU AI Act (fully enforced August 2026), GDPR, US state laws like BIPA, and PIPL. It covers crucial areas such as lawful basis, DPIA, template storage, bias testing, and liveness detection for legal professionals dealing with biometric access control, payment flows, or identity verification services.

Over 1,000 AI-related patent lawsuits have been filed globally in the past five years, with the US accounting for 65% of cases. Deep learning algorithms and natural language processing (NLP) are involved in 70% and 40% of these disputes, respectively, indicating key areas of legal contention and innovation.

A federal judge found that Anthropic's destructive scanning of millions of physical books to train its Claude AI model was "transformative" and qualified as fair use. The ruling emerged from a copyright class-action lawsuit where a $1.5 billion settlement was approved, revealing Anthropic's internal discussions about the cost-effectiveness of using pirated books and the physical destruction process.

The 9th Circuit vacated a preliminary injunction against Perplexity AI, ruling that the user — not Perplexity — likely "accessed" Amazon's computers when directing an AI assistant. The decision narrows the scope of Computer Fraud and Abuse Act (CFAA) and California CDAFA for AI developers, though it is limited to the specific facts and statutes.

A study found that 76% of widely downloaded open source AI models use training data with incompatible or more restrictive licenses than the model's declared license. This creates significant legal and commercial risks for companies deploying these models, including potential copyright infringement or forced open-sourcing of proprietary models.