anthropic launches amazon while meta expands its platform

Key Takeaways

• Anthropic accidentally leaked over 500,000 lines of its Claude Code AI tool's source code twice due to human error, not a security breach.
• The Anthropic leaks revealed unreleased features like Kairos (a persistent background daemon), an AI 'dream' process, and 'Undercover mode'.
• Anthropic issued over 8,000 copyright takedown requests on GitHub to remove the leaked Claude Code.
• Oracle is laying off thousands (reportedly 10,000) of employees globally to fund significant investments in AI infrastructure and data centers.
• Meta is implementing an AI-powered Risk Review program to accelerate and improve the identification of privacy, safety, and security concerns in product development.
• Amazon's new AI chat ads are collecting consumer data but are not yet driving significant sales and can sometimes provide inaccurate information.
• Mark Cuban advises businesses to adopt AI technologies, stating that companies not using large language models risk falling behind.
• Adversa AI won an award for its innovative Agentic AI Security platform, which helps test AI systems against attacker techniques.
• AI is being used to generate fake grassroots opposition, flooding public agencies with deceptive comments using real people's identities.
• A student startup, Drive AI, helps small and mid-sized businesses automate administrative tasks using AI, while India's IAIRO aims to bridge the 'lab-to-market gap' for AI products.

Anthropic leaks 500,000 lines of its own source code

Anthropic accidentally released about 500,000 lines of its own source code for its Claude Code AI tool. This leak happened due to a packaging error caused by human mistake, not a security breach. The leaked code revealed details about unreleased features like a persistent assistant and remote capabilities. While no sensitive customer data was exposed, the leak gives competitors a look at Anthropic's development plans and internal workings.

Anthropic's Claude Code source code leaks again

Anthropic's source code for its Claude Code AI tool was leaked for the second time, revealing internal blueprints and unreleased features. The company confirmed the leak was due to human error in packaging, not a security breach, and stated no sensitive customer data was exposed. Programmers have been analyzing the code, finding details about Claude's operations and potential future capabilities. This leak occurs at a critical time for Anthropic, potentially impacting its business and competitive standing.

Anthropic races to contain Claude AI code leak

Anthropic is working to control the impact of a leak of its Claude Code AI agent's source code. The company has used copyright takedown requests to remove over 8,000 copies from GitHub. Anthropic stated that the leak, caused by human error in packaging, did not expose customer data or reveal the core mathematics of its AI models. This incident follows another recent data exposure and raises concerns about the AI company's security practices.

Claude Code creator says leak was human error, no one fired

Boris Cherny, creator of Claude Code, stated that the recent source code leak was due to human error in the deployment process, not a security breach. He explained that a manual step was missed, leading to the accidental inclusion of internal code in a public update. Cherny emphasized that no sensitive customer data was exposed and that the focus is on improving automation to prevent future mistakes. He also confirmed that no employees were fired over the incident, reinforcing a culture of learning from process errors.

Anthropic accidentally leaks thousands of lines of code

Anthropic unintentionally released internal source code for its Claude Code AI assistant, raising security questions for the safety-focused company. The leak, involving about 512,000 lines of code, was attributed to a packaging error and human mistake, not a breach. This is Anthropic's second security lapse in days, following a previous exposure of internal files. Experts worry about potential vulnerabilities and the advantage this gives competitors.

Claude Code leak reveals Anthropic's future plans

The recent leak of Anthropic's Claude Code source code has revealed details about its future development roadmap. Observers found references to hidden features like Kairos, a persistent background daemon with a file-based memory system for continuous operation. The code also includes prompts for an AI 'dream' process to manage memories and an 'Undercover mode' to prevent information leaks. These discoveries offer a glimpse into Anthropic's strategy for enhancing Claude Code's capabilities.

Anthropic accidentally releases Claude AI agent source code

Anthropic unintentionally released internal source code for its Claude AI agent in a packaging error caused by human mistake. The company stated that no sensitive customer data or credentials were exposed. This is the second security lapse from Anthropic in a week, following a leak of internal files. The incident has sparked community discussion about the AI agent's workings and raised concerns about potential security vulnerabilities.

Anthropic code leak exposes AI security before IPO

Anthropic accidentally published over 500,000 lines of its Claude Code source code, revealing its security architecture just months before a potential IPO. The leak, discovered on March 31, included features like 'Undercover Mode' and codenames for upcoming models. This second data leak in a week has raised concerns among investors and competitors. The rapid replication of the code by developers highlights the fragility of proprietary AI technology.

Oracle cuts thousands of jobs to fund AI investments

Software giant Oracle is laying off thousands of employees globally to increase spending on AI infrastructure. Affected workers in the US, India, and other regions reported receiving termination notices simultaneously. These cuts come as Oracle invests heavily in AI, competing with cloud providers like Amazon. The company's stock has fallen this year, and analysts suggest the layoffs are a strategic move to improve cash flow for AI expansion.

Oracle reportedly cuts 10,000 jobs for AI investments

Oracle is reportedly conducting mass layoffs, with estimates suggesting around 10,000 positions have been cut across various divisions. Reports indicate these job reductions are intended to fund significant investments in AI, including data centers. Some affected employees may be replaced by AI, a trend that raises concerns about the future job market. This follows earlier reports of Oracle facing financial strain and underperforming its tech peers.

Adversa AI wins award for innovative AI security platform

Adversa AI has won the 'Most Innovative Agentic AI Security' Platform award at the Global InfoSec Awards during RSA Conference 2026. The company was recognized for its advancements in Continuous AI Red Teaming and Agentic AI security. Adversa AI's platform helps organizations test AI systems against real attacker techniques, detect vulnerabilities, and validate agent behavior. This award highlights the growing importance of securing AI agents as they are increasingly deployed in enterprise environments.

Game theory falls short in AI trading markets

Game theory struggles to fully explain the complexities of AI-driven trading markets, according to AI academic Bo An. He explained that financial markets have too many shifting strategies and payoffs to fit traditional game theory models. An believes adaptive strategies based on data are more effective than relying on game theory. He also noted that AI currently excels at recognizing patterns over causation, unlike human intelligence.

AI campaigns flood agencies with fake comments

Artificial intelligence is being used to create fake grassroots opposition, flooding public agencies with deceptive comments. Organizations are submitting emails and comments to regulators using real people's identities without their consent, often to oppose environmental measures. This tactic, employed by lobbying firms and fossil fuel companies, undermines democratic input processes. Investigations are needed to uncover how AI was used, where the identities came from, and who funded these deceptive campaigns.

Mark Cuban urges businesses to adopt AI

Entrepreneur Mark Cuban advised D-FW business owners to adopt AI technologies, warning that those not using large language models are falling behind. Speaking at the Convergence AI Dallas conference, Cuban compared AI adoption to the early days of personal computers and the internet. He emphasized that AI can act as a powerful tool for learning and productivity, creating a divide between companies that embrace it and those that don't. Cuban predicts that companies failing to integrate AI will struggle to survive.

Meta uses AI to speed up product risk reviews

Meta is deploying an AI-powered Risk Review program to accelerate and improve the identification of potential privacy, safety, and security concerns during product development. The AI automates tasks like prefilling documentation and scanning proposals, strengthening human judgment rather than replacing it. This allows Meta to apply safeguards more consistently and monitor outcomes continuously. The company believes this AI evolution will provide better protections for its billions of users.

Amazon's AI chat ads gather data but few sales

Amazon's new AI chat ads, currently in a limited beta, are generating valuable data about consumer interests but are not yet driving significant sales. Marketers testing the 'Sponsored Products' and 'Sponsored Brands' ads report that the AI chatbot sometimes provides inaccurate or irrelevant information. While Amazon refines the product, some marketers remain optimistic about the potential of AI chat ads for personalized recommendations and real-time customer engagement in e-commerce.

Student startup Drive AI helps businesses use AI

Clemson University students Reid Turner and Danika Pfleghardt founded Drive AI to help small and mid-sized businesses leverage artificial intelligence. The startup streamlines operations by automating repetitive administrative tasks, allowing employees to focus on higher-value work. Drive AI emerged from a classroom conversation and grew through university networking and mentorship. They aim to bridge the gap by making powerful AI tools accessible and practical for smaller organizations.

India has AI talent but lacks products, say founders

The Indian Artificial Intelligence Research Organisation (IAIRO) founders state that India produces skilled AI talent but struggles to translate research into scalable products. They aim to address this 'lab-to-market gap' by creating a national capability similar to ISRO for space. IAIRO focuses on converting prototypes to products and integrating startups, emphasizing foundational data infrastructure and smaller, specialized AI systems for decision support rather than just large language models.

Sources

• Anthropic leaked 500,000 lines of its own source code
• Source Code for Anthropic's Claude Code Leaks at the Exact Wrong Time
• Anthropic Races to Contain Leak of Code Behind Claude AI Agent
• Claude Code Leak Was ‘Human Error’, No One Was Fired: Claude Code Creator Boris Cherny
• Anthropic accidentally leaked thousands of lines of code
• Here's what that Claude Code source leak reveals about Anthropic's plans
• Anthropic Accidentally Releases Source Code for Claude AI Agent
• Anthropic Source Code Leak Exposes AI Security Logic Before $350B IPO
• Oracle cuts thousands of jobs amid intensified AI investment push
• Oracle believed to have cut 10,000 positions across multiple divisions as mass layoffs begin to fuel AI investments — company is reportedly reducing headcount to fund data centers
• Adversa AI Wins "Most Innovative Agentic AI Security" Platform at Global InfoSec Awards During RSA Conference 2026
• Why game theory falls short in AI-driven trading market
• Contributor: Investigate the AI campaigns flooding public agencies with fake comments
• Mark Cuban tells D-FW business owners to get with the AI program
• Meta Deploys AI to Accelerate and Enhance Risk Review During Product Development
• Amazon’s AI Chat Ads Yield Data but Few Sales
• From a Classroom Conversation to a Growing AI Startup
• India is producing AI talent, but not products, say IAIRO founders