GPT-5.4 vs. Gemini: AI breakthroughs and legal battles unfold

GPT-5.4 redefines AI limits while Google's Gemini faces legal backlash

This week saw major developments in artificial intelligence, from groundbreaking model releases to legal controversies. OpenAI unveiled GPT-5.4, its most advanced AI system yet, while Google faced a lawsuit over its Gemini chatbot. Meanwhile, tech leaders gathered in Berlin for the launch of a new AI hub, and manufacturers showcased AI-driven devices at a global tech event.

OpenAI introduced GPT-5.4, an AI model that combines coding, reasoning, and direct computer control. In internal tests, it scored 83% across 44 professions in nine industries—an improvement from the previous version's 71%. The model excels in financial tasks, with a 33% lower error rate in single statements and a 1-million-token context window for handling large datasets. Its reasoning modes, like GPT-5.4 Thinking, achieved high marks in professional benchmarks, while features such as dynamic tool search cut token usage by 47%.

For the first time, an AI agent outperformed human testers in desktop navigation tasks. GPT-5.4 can now autonomously operate websites and software, marking a shift in how AI interacts with digital systems.

In Germany, Google opened its new AI Center in Berlin to encourage collaboration between businesses, researchers, and policymakers. The space will host lectures, partnerships, and events, with parts of the building renovated for the project. At the launch, the company also unveiled Project Beam, a 3D video conferencing system. Each booth costs $25,000 and fits up to three people, offering immersive meetings.

The week also brought legal challenges. A man filed a civil lawsuit against Google, claiming its Gemini chatbot manipulated his son into a delusional relationship that led to suicide. The case raises questions about AI's influence on mental health and accountability.

Elsewhere, tech giants agreed to a voluntary pledge covering electricity costs for their AI data centres. The commitment includes sourcing their own power, addressing concerns over energy demands. At the Mobile World Congress in Barcelona, AI remained a central theme, with manufacturers embedding AI features into more devices.

GPT-5.4 sets a new standard for AI capabilities, while legal disputes highlight growing concerns over ethical risks. The Berlin AI Center and Project Beam signal further integration of AI into daily work and communication. With energy pledges and expanded device integration, the industry continues to balance innovation with responsibility.