Pulse on AI: Memory, Deepfakes, and the Data Center Boom: Navigating AI’s Growth Pains

Today’s Pulse on AI dives into OpenAI’s culture shift toward growth and ads—potentially leveraging ChatGPT’s Memory—plus Sora’s moderation challenges and Sam Altman’s warning about “strange or scary moments.” We unpack a BBC-led study finding major inaccuracies in AI news summaries, the massive AI data center build-out and its environmental trade-offs, and Xataka’s week-long test of the Hypershell X Pro exoskeleton. We cover pragmatic career strategies for the AI era, how to tell durable ARR from hype in AI startups, a toy study on optimal model size vs. data under fixed compute, the ransomware confidence gap amid AI-driven attacks, decentralized efforts to detect deepfakes, Germany’s push to level rules for platforms and media, and how Spotify, YouTube Music, Apple Music, and TIDAL use AI to surface new music. Three takeaways: prioritize trust and transparency, favor practical AI with measurable ROI, and chase efficiency across models and infrastructure.

Sources:

• OpenText 2025 Ransomware Survey: Confidence Meets AI Threats
• Can Decentralised Tech Restore Truth in the Age of Deepfakes and AI?
• Sam Altman: AI will cause “strange or scary moments”
• Cómo usar Spotify, YouTube Music o Apple Music para descubrir música nueva
• BBC probe finds AI chatbots mangle nearly half of news summaries
• The Splendor And Misery Of ARR Growth
• ChatGPT's memory could turn personal details into ads OpenAI CEO Altman once called dystopian
• Choosing the Best Model Size and Dataset Size under a Fixed Budget for LLMs
• Deutschland gegen die Algorithmen: KI und Plattformen sollen an die Leine
• How Data Centers Actually Work
• El Hypershell X Pro se presenta como el exoesqueleto más avanzado. Lo hemos probado una semana y hay un dilema evidente
• 5 ways ambitious IT pros can future-proof their tech careers in an age of AI