Anthropic code leak raises security concerns

04/26, AI Models & Tools, Global

02/04/2026

Anthropic code leak raises security concerns

Anthropic accidentally exposed the source code of its popular programming tool, Claude Code. Around 500,000 lines of code and roughly 1,900 files were published through a misconfigured npm package. The incident was caused by human error, not a cyberattack. No customer data was affected, but internal structures and unreleased features became visible, raising security concerns across the industry.

This is the second major incident in just a few days, following the accidental exposure of internal files for an upcoming AI model. Security experts warn that competitors could use the leaks to reverse-engineer Anthropic’s technology and highlight vulnerabilities in protecting sensitive development data.

Read the article

Updates from the world of artificial intelligence

Recent News

26. Mai 2026

Google massively expands AI offerings

At its I/O 2026 developer conference, Google introduced several new AI products. These include Gemini Spark, an autonomous assistant designed to manage tasks across various Google services, and Gemini Omni, a multimodal tool for creating and editing videos.

26. Mai 2026

Controversy over Palantir deal

A political and public dispute has erupted in London after Mayor Sadiq Khan blocked a £50 million contract between the Metropolitan Police and US data analytics company Palantir.

26. Mai 2026

Nvidia says AI is ready for mainstream

Nvidia is trying to convince increasingly skeptical investors that artificial intelligence has moved beyond the hype phase and is entering both the consumer and business mainstream.

26. Mai 2026

SpaceX with AI bets and Spaceship dreams

The IPO filing presents SpaceX as far more than just a rocket company. It ties the company’s future to artificial intelligence, spaceship development, and Elon Musk’s long-term ambitions for humanity.

26. Mai 2026

Many exaggerate their AI skills

A survey shows that nearly two-thirds of employees admit to overstating their AI skills at work in order to appear more competitive. Many feel pressured to keep up with the rapid adoption of AI in the workplace.

26. Mai 2026

EU simplifies strictest AI law

The EU has revised and simplified its comprehensive AI regulation, known as the AI Act. The goal is to reduce bureaucratic hurdles and make it easier for companies to comply with the rules, while maintaining the law’s risk-based approach.

19. Mai 2026

Laserfiche introduces new AI agents

Laserfiche is rolling out AI agents that can act autonomously within the platform on behalf of users, executing complex, multi-step workflows based on instructions in natural language.

19. Mai 2026

Europe aims for digital sovereignty

Europe wants to reduce its dependence on US technology. But weaknesses in cloud infrastructure, identity management and public sector procurement make this difficult.

19. Mai 2026

Training AI to be more skeptical

An experiment shows that ChatGPT’s hallucinations can be significantly reduced when the model is prompted to distrust its own answers. With appropriate prompts, the AI acts like a “critical reviewer,”

19. Mai 2026

Germany backs European AI

Germany’s domestic intelligence agency, the BfV, has chosen French company ChapsVision and its ArgonOS platform instead of Palantir. This sends a clear signal for greater European digital sovereignty in sensitive security infrastructure.

19. Mai 2026

AI data centers face growing resistance

The boom in AI data centers is facing increasing resistance. While tech companies continue expanding infrastructure for AI models and AI tools, many communities are pushing back against rising electricity and water consumption, noise, land use and lack of transparency.

19. Mai 2026

ArXiv tightens AI rules

Research platform arXiv will suspend authors for one year if submitted papers contain clear signs of unchecked AI-generated content. The policy targets researchers who rely too heavily on AI without verifying facts, claims, or citations.

19. Mai 2026

Court questions AI-powered facial recognition

A German district court has rejected an arrest warrant that was based largely on a match generated by an AI-powered facial recognition system. According to the judges, algorithmic results alone are not sufficient grounds for detention.

12. Mai 2026

Germany pushes sovereign AI forward

STACKIT, the Schwarz Group’s cloud provider, and Cologne-based neuland.ai are building a German-operated AI architecture for companies with strict compliance requirements. The offering combines STACKIT’s cloud infrastructure with the neuland.ai HUB.

12. Mai 2026

AI toys raise safety concerns

AI-powered toys are spreading fast, from plush companions to child-friendly robots. But clear rules are still missing, and many products use chatbot models designed for adults.

12. Mai 2026

OpenAI launches three new audio models

OpenAI has introduced three new real-time audio models: GPT-Realtime-2 for conversations between humans and AI, GPT-Realtime Translate for live translation of human conversations, and GPT-Realtime Whisper for low-latency transcription.

12. Mai 2026

EU temporarily simplifies AI rules

EU member states and the European Parliament have provisionally agreed to simplify AI rules under the “Digital Omnibus” package. Some key obligations will be postponed to reduce legal uncertainty and avoid double regulation.

12. Mai 2026

Google Workspace introduces five AI upgrades

Google is expanding Gemini across Workspace with five AI upgrades designed to make everyday work faster and more connected. These include natural-language data analysis in Sheets, interactive visualizations, and reusable automations.

12. Mai 2026

SAP is building a European AI center of excellence

SAP plans to acquire the German start-up Prior Labs and invest around $1.16 billion to build a leading European AI research center focused on enterprise data. Prior Labs was founded in Freiburg just 18 months ago and specializes in “tabular foundation models.”

12. Mai 2026

Dutch GPT-NL enters the real world

GPT-NL, developed in the Netherlands, is now being tested in real-world settings. It is a national AI model designed to offer a European alternative to systems from Silicon Valley.

5. Mai 2026

Pentagon signs contracts for military AI projects

The Pentagon has signed agreements with seven AI companies, including SpaceX, OpenAI, Google, Nvidia, Microsoft, Amazon Web Services, and Reflection. They are set to work on classified military projects.

5. Mai 2026

Measures to reduce AI’s cognitive costs

AI is meant to reduce routine work, but research suggests it can also increase workload. Simple, recurring tasks may decrease, while demanding analysis, assessment, and oversight can grow significantly.

5. Mai 2026

Cohere bets on sovereign AI

Cohere is taking over German company Aleph Alpha to build a Canadian-German AI company focused on sovereign systems for enterprise use. The transaction is backed by Schwarz Group with €500 million in structured financing.

5. Mai 2026

Europe’s rising AI talent hubs

Europe is becoming increasingly important as a destination for AI professionals, even though the United States and India continue to dominate globally. A new study finds that Ireland, Germany, and the Netherlands are getting especially attractive.

5. Mai 2026

Better prompts for AI image editing

Powerful AI photo editing today depends less on the tool itself and more on clear instructions. On platforms such as ChatGPT, Gemini, Copilot, and Meta AI, the best results come from precise prompts about lighting, texture, mood, background, and realism.

5. Mai 2026

WordPress tool cuts web crawling costs

The British nonprofit organization The Chancery Lane Project has released a new open-source plugin for WordPress that could significantly reduce the cost and energy used when AI agents crawl websites.

5. Mai 2026

Swiss banks assess AI risks from Mythos

The Swiss financial sector is closely watching Anthropic’s powerful AI model Mythos, but does not yet see it as an immediate crisis. Mythos can reportedly detect unknown software bugs quickly.

28. April 2026

AI needs integration, not hype

Many companies still see AI as a silver bullet that will cut costs overnight and transform their business. Instead of chasing a single breakthrough tool, companies should gradually embed AI into their existing processes as invisible infrastructure.

28. April 2026

Google bets billions on Anthropic

Google is significantly expanding its AI investments with a potential commitment of up to $40 billion in Anthropic, the company behind Claude. The deal includes an immediate $10 billion investment, along with another $30 billion tied to specific performance targets.

28. April 2026

Europe’s industrial challenges in the AI era

Europe risks losing industrial leaders like Siemens if it fails to balance AI innovation with competitiveness. Stricter regulation, higher energy costs, and slower decision-making are pushing manufacturers toward the US and China.

28. April 2026

Anthropic expands into legal services

The international commercial law firm Freshfields has entered into a multi-year partnership with Anthropic to develop AI-powered legal workflows and roll out Claude across the firm globally.

28. April 2026

GPT-5.5 – one step closer to an „AI Super App“

OpenAI has introduced GPT-5.5 and sees the model as an important step toward building a “super app.” According to OpenAI president Greg Brockman, the company is moving significantly closer to its vision of a more “agentic and intuitive” way of computing.

28. April 2026

A billion-dollar bet against LLMs

Yann LeCun, former Chief AI Scientist at Meta, is taking a very different approach in the AI race. His startup AMI Labs has raised $1.03 billion in seed funding and is already valued at $3.5 billion.

28. April 2026

Musk versus Altman

Elon Musk and Sam Altman are heading into a high-stakes legal battle that could significantly reshape OpenAI. Musk accuses Altman and OpenAI of abandoning the company’s original nonprofit mission in favor of a profit-driven model.

28. April 2026

Bosch advances autonomous driving

Bosch is bringing autonomous driving one step closer to the mass market with a new Level 3 system based on AI and by-wire technology. Unlike the previous system, drivers can now take their hands off the wheel and their eyes off the road in certain situations.

21. April 2026

The reasons behind Anthropic’s peak valuation

Anthropic, the AI company behind the Claude models, is experiencing a surge in investor interest. According to reports, venture capital firms are considering valuations of up to $800 billion.

21. April 2026

Berlin instead of Zurich for AI startup

Adrian Locher, founder of the AI venture studio Merantix, deliberately chose Berlin over Zurich when building his company, seeing it as a better environment for an AI ecosystem.

21. April 2026

Why Artificial Intelligence shows bias

AI systems are trained on human data and built on human assumptions. As a result, they exhibit predictable and systematic biases when evaluating people. These systems don’t truly “understand” individuals.

21. April 2026

China is rapidly closing the AI gap

The latest Stanford AI Index Report shows that China has significantly narrowed the gap with the United States in artificial intelligence. In key performance indicators, the two countries are already nearly on par.

21. April 2026

Merz calls for eased EU rules on industrial AI

German Chancellor Friedrich Merz has called for less strict EU regulations on industrial artificial intelligence. At the trade fair in Hannover he was arguing that current rules are too restrictive for businesses.

21. April 2026

OpenAI facing key questions about the future

OpenAI is facing important strategic and fundamental questions about the future of artificial intelligence, including the development of sustainable business models and how to address long-term societal risks.

21. April 2026

Anthropic’s Mythos and what to do next

Growing concerns about cybersecurity threats should not lead to exaggerated “vulnocalypse” scenarios. Large language models are increasingly capable of detecting software vulnerabilities, but they do not automatically trigger uncontrollable disasters.

14. April 2026

Potential German-Canadian AI Merger

The Canadian AI company Cohere is in advanced talks to merge with the German firm Aleph Alpha, aiming to create a transatlantic player in artificial intelligence. The potential deal is supported by the German government.

14. April 2026

AI governance protects corporate profits

According to IBM, robust AI governance is essential for protecting corporate margins as AI increasingly becomes a core layer of business infrastructure. Given the typical technology lifecycle governance is becoming more important at every stage.

14. April 2026

AI struggles with soccer betting

A recent study shows that advanced AI models perform poorly when betting on soccer matches and often incur losses, even though they have access to extensive historical data.

14. April 2026

Desire for AI independence vs. actual AI usage

A recent Bitkom survey reveals a significant gap between digital ambitions and real-world usage in Germany. While 99% of respondents consider digital independence important, and 93% believe Germany is overly dependent on other countries.

14. April 2026

Europe leads in AI research, but lags behind in commercialization

Europe achieves outstanding results in research across artificial intelligence, drones, robotics, quantum technologies, and cybersecurity, establishing itself as one of the world’s leading scientific regions.

14. April 2026

Setback for the UK’s AI ambitions

OpenAI has put its involvement in the UK’s £31 billion “Stargate” AI mega-project on hold for now. The company cites high energy costs and regulatory uncertainty as the main reasons.

9. April 2026

Shadow AI calls for new control solutions

KiloClaw is a new platform designed to help companies better manage and monitor autonomous AI agents. At its core is the growing issue of “shadow AI,” where employees use AI tools on their own without official approval or oversight from IT.

9. April 2026

Google Vids gets major AI upgrade

Google has rolled out a significant update for its tool Google Vids. The update integrates models like Veo 3.1 for high-quality video creation, Lyria 3 for custom music, and directable AI avatars.

Anthropic accidentally exposed the source code of its popular programming tool, Claude Code. Around 500,000 lines of code and roughly 1,900 files were published through a misconfigured npm package.