OpenAI has introduced a new AI model called "Astra for Law: GPT-6 Astra," specifically tailored for legal professionals. This model is designed to assist lawyers in legal research, argument development, and document drafting by providing access to a dedicated legal research index. This index allows searches across U.S. case law, regulations, procedural rules, and administrative decisions, covering over 230 million web pages. To ensure compliance with legal privacy standards, OpenAI is implementing measures such as "secure access programs" that limit access to qualified lawyers and their teams. Additionally, the model includes "zero data retention," meaning user data is not stored, and by default, human oversight is excluded when using ChatGPT Enterprise. OpenAI is collaborating with major law firms like Cooley, Ropes & Gray, and Sullivan & Cromwell to integrate Astra for Law into their workflows. These partnerships aim to develop tools that enhance efficiency and accuracy in legal practice. The company is also launching 26 plugins developed by legal technology partners such as Thomson Reuters and iManage, as well as nine community plugins created by legal professionals. OpenAI has emphasized its long-term commitment to the legal field, promising ongoing improvements to the model based on feedback from lawyers and legal technology experts. In November 2025, U.S. courts began encountering a surge in "slopified" legal documents—AI-generated or heavily AI-assisted texts that distorted logic, invented case law, and obscured legal language. This trend sparked concern among legal professionals, who criticized the premature and uncritical adoption of AI tools. In February 2026, a lawyer named Steven Feldman faced severe consequences after accumulating multiple AI-generated errors in his arguments, despite prior warnings from the court. The case highlighted the risks of over-reliance on AI and raised questions about the ethical responsibilities of legal professionals in an age of algorithmic tools. Earlier, in June 2023, lawyer Steven Schwartz was fined $10,000 and required to complete ethics training after citing six fabricated cases generated by ChatGPT. The judge described the practice as "unprecedented" and "deceptive," warning of the broader risks of AI in legal work. Similarly, in a recent case, lawyer Stephen Aarons was fined $5,000 and found guilty of contempt of court for using AI-generated fabricated testimony in a murder case. Aarons claimed he had not anticipated the AI would invent facts, but the court was unconvinced, citing his lack of remorse and referring him to a disciplinary committee. These cases reflect a growing trend of legal professionals facing sanctions for misusing AI tools. As the use of AI in legal work continues to expand, OpenAI is part of a growing number of companies developing AI models tailored for the legal field. Microsoft introduced "Legal Agent" in Word, Anthropic launched "Claude for Legal," and Google announced "Gemini Enterprise for Legal." These developments signal an increasing recognition of the need for specialized AI tools in law, even as the profession grapples with the ethical and practical challenges of integrating such technologies.