openai launches anthropic while chatgpt expands its platform

Key Takeaways

• Sullivan & Cromwell apologized to Judge Martin Glenn for AI hallucinations causing inaccurate citations in a Prince Global Holdings bankruptcy filing.
• OpenAI committed $10 million in API credits to help banks defend against AI-driven threats via its GPT-5.4-Cyber model.
• Elastic launched MCP Apps to embed security and observability tools directly into third-party AI chat interfaces like Claude Desktop.
• Computer science students at the University of Arkansas worry about job prospects following a reported decline in employment for ages 22-25 in tech sectors.
• Anthropic is suing music publishers over copyright, arguing fair use for training its Claude chatbot on song lyrics.
• Tencent's QClaw AI agent product generated 99% of its code autonomously and is now available internationally via email and messaging apps.
• Travel experts warn that AI booking tools may use outdated data, advising users to verify prices and availability before purchasing.
• EDB reports that many enterprises ask the wrong questions about AI, focusing on technical models rather than aligning with business strategy.
• Zynt raised 424.4k euros to fund its AI-powered B2B sales intelligence platform that identifies buying triggers using machine learning.
• AI is being integrated into hospitality to improve wine recommendations and guest satisfaction without replacing human staff.

Law Firm Apologizes for AI Errors in Court Filing

Sullivan & Cromwell filed an emergency motion asking a judge not to sanction them for AI-generated errors in a legal document. The firm admitted that artificial intelligence hallucinations created inaccurate case citations and other mistakes in a filing for the Prince Global Holdings bankruptcy case. Andrew Dietderich, a co-head of the firm, apologized to Judge Martin Glenn and thanked Boies Schiller Flexner for catching the issues. The firm stated it did not follow its own AI policies and review processes during the preparation of the motion.

Sullivan & Cromwell Apologizes for AI Errors in Court Filing

Reuters reported that Sullivan & Cromwell apologized to a federal judge for submitting a court filing with inaccurate citations generated by artificial intelligence. The errors, known as AI hallucinations, included fabricated case citations and misquoted laws in a document dated April 18. Andrew Dietderich, a partner at the firm, wrote to Judge Martin Glenn of the U.S. Bankruptcy Court in Manhattan to express regret for the mistakes. The firm admitted that its secondary review process failed to identify the inaccuracies generated by the AI tools.

Law Firm Apologizes to Judge for AI Errors in Filing

A lawyer from Sullivan & Cromwell apologized to a bankruptcy judge for filing documents with incorrect case citations caused by artificial intelligence. Andrew Dietderich, co-chair of the firm's restructuring practice, told the judge that the firm did not properly review legal citations and did not follow its protocols. The apology occurred on April 18 after the error-ridden documents were filed ten days earlier in the Southern District of New York. The firm provided a red-line version of the document to correct the errors, which involved fixing entire sentences and citation details.

Top Law Firm Admits to AI Errors in Court Filing

An elite Wall Street law firm apologized to a federal judge for submitting a court filing full of errors created by artificial intelligence. Andrew Dietderich, a partner at Sullivan & Cromwell, wrote to Judge Martin Glenn on April 18 to express deep regret over the situation. The A.I.-generated errors included fabricated case citations and inaccurate titles of articles in a motion filed in U.S. Bankruptcy Court in Manhattan. The firm is one of the oldest and most prestigious law firms in the country and represents President Trump in several appeals.

Elite Law Firm Admits to AI Errors in Court Filing

Sullivan & Cromwell admitted to hallucinations in its use of artificial intelligence and apologized to a judge for software-driven errors in a bankruptcy case. The firm acknowledged that its AI tools produced inaccurate information, leading to mistakes in a legal filing for the Prince Global Holdings case. The incident has raised questions about the reliability of AI in high-stakes legal work and the potential consequences of over-reliance on technology. The firm has pledged to review its AI practices and implement stricter controls to prevent similar incidents in the future.

Elastic Brings Security Tools Inside AI Chats

Elastic has delivered its first embedded AI experiences for observability and security inside third-party AI tools like Claude Desktop. The company created MCP Apps that allow teams to manage threat detection and system diagnosis without leaving the conversation interface. Mandy Andress, CISO of Elastic, explained that these apps bridge the gap between automated detection and manual hunting by bringing security data directly into a single interface. The Security MCP App enables analysts to triage alerts, run queries, and investigate threats through interactive views rendered directly in the conversation.

Elastic Launches AI Tools Inside Third-Party Chat Apps

Elastic launched its first embedded AI experiences for observability and security inside third-party AI tools to help teams act on data where they work. The company's MCP Apps support security and observability workflows in a live AI-native interface that users can explore, filter, and act on without switching tools. Ken Exner, chief product officer at Elastic, stated that customers are increasingly working inside AI-native environments and need workflows integrated directly into those tools. The Observability MCP App helps engineers move from detection to root cause analysis through interactive views of distributed traces and service dependencies.

AI Improves Wine Sales Without Replacing Staff

AI is being integrated into hospitality to improve wine recommendations on the restaurant floor and enhance guest satisfaction without removing the human element. Rachel Wilson, general manager of the Napa Valley Marriott, said personalization is central to the approach, helping match the right glass to the right guest. The technology reduces decision friction for guests and improves product discovery while allowing staff to offer targeted suggestions without aggressive upselling. This approach augments servers and sommeliers to help them sell better rather than forcing them to push products.

OpenAI Secures Banks with New AI Security Tools

OpenAI is scaling its GPT-5.4-Cyber model and Trusted Access for Cyber program to help banks defend against AI-driven threats. The company committed US$10m in API credits through its Cybersecurity Grant Program to support defenders in the industry. Major financial institutions including Bank of America, BlackRock, JPMorgan Chase, and Goldman Sachs have already adopted the Trusted Access for Cyber program. GPT-5.4-Cyber is engineered specifically for defensive cybersecurity use cases within financial and enterprise environments to detect and prevent threats.

Arkansas Students Worry About AI Impact on Jobs

Computer science students at the University of Arkansas expressed concerns about job prospects in a growing AI market. Caleb Moline, a sophomore, said he does not even know what entry-level means anymore because the bar for experience has risen significantly. A recent report from Anthropic showed a decline in employment for workers aged 22 to 25 in computer and math sectors affected by AI. While some students feel economic factors drive the hiring slowdown, others believe companies are using AI as a facade for a struggling economy.

Zynt Raises Funding for AI Sales Intelligence Tool

Warsaw-based company Zynt raised 424.4k euros in pre-Seed funding to sharpen its AI-powered B2B sales intelligence platform. The funding was secured from Poland-based fund 24Ventures and a group of angel investors. Zynt helps uncover buying triggers in every industry by aggregating and analyzing signals from online content, social media activity, and official business registries. The platform focuses on identifying when a potential customer is most likely to buy rather than just who to contact. Co-founder Cezary Raszel noted that the system uses machine learning and natural language processing to learn which signals matter within a sales pipeline.

Tencent Launches AI Agent Product for Global Users

Tencent opened an international beta for its AI agent product QClaw, which allows users to deploy and operate AI agents through email and messaging apps. The product is built on the OpenClaw open-source framework and supports multiple large language models with custom model integration through API keys. QClaw includes long-term memory features and a security module called Gateway to monitor AI agent activity in real time. The international version was developed in five days and 99% of its code was generated autonomously by QClaw itself.

Travel Experts Warn About AI Booking Errors

Sally French, a travel expert with NerdWallet, advised travelers to use AI as a baseline for searches but take the information with a grain of salt. She warned that AI models might use data trained in 2024, so exact prices or hotel availability from AI could be outdated. French recommended that travelers confirm and research whatever information they get from AI about travel before booking. She added that providing detailed inputs to AI, such as group size and budget, helps generate better personalized recommendations for flights and hotels.

Anthropic Sues Over AI Training of Music Lyrics

Anthropic asked a California federal court to rule in its favor in a copyright lawsuit brought by music publishers Universal Music Group, Concord, and ABKCO. The company argued it made fair use of song lyrics to train its AI-powered chatbot Claude for understanding human language. Anthropic stated that its training was transformative and intended to help progress in science, business, and education. The company asked Judge Eumi Lee for summary judgment, arguing that copyright law encourages such transformative technology rather than allowing copyright holders to veto it.

Experts Say Enterprises Ask Wrong Questions About AI

EDB observes that many enterprises are still asking the wrong questions when trying to implement AI agents in their organizations. Leaders often focus on technical choices like which model to use before aligning on business priorities that AI should serve. The report suggests that organizations should figure out how to use AI to support their existing business strategy rather than creating a new AI strategy from scratch. Companies that recognize data sovereignty as mission-critical are winning because they own their data and AI capabilities.

Sources

• Sullivan & Cromwell Files Emergency ‘Please Don’t Sanction Us For All These AI Hallucinations’ Letter
• Sullivan & Cromwell law firm apologizes for AI 'hallucinations' in court filing
• Sullivan & Cromwell Apologizes to Judge for AI Hallucinations
• A.I. ‘Hallucinations’ Created Errors in Court Filing, Top Law Firm Says
• Elite law firm Sullivan & Cromwell admits to AI ‘hallucinations’
• Elastic puts threat hunting and system diagnostics inside AI chats
• Elastic Delivers First Embedded AI Experiences for Observability and Security Inside Third-Party AI Tools
• AI Enhances Wine Sales Without Replacing Staff
• How OpenAI's Secure AI Shields Financial Giants From Threats
• UA engineering students navigate an AI-affected job market
• Warsaw’s Zynt raises €424.4k pre-Seed to sharpen AI-powered B2B sales intelligence
• Tencent opens international beta for AI agent product QClaw
• AI travel booking: What works and what to watch out for
• Anthropic seeks pivotal court win in music publisher lawsuit over AI training
• Billions of AI Agents Are Coming to Work: Yet Most Enterprises Are Still Asking the Wrong Questions