ENFR
8news

Tech • IA • Crypto

TodayTopicsVideosCryptoArchivesFavorites

GPT-5.6 Sol, IBM -25%, EU vs Meta DSA, Anthropic Ad

AIWednesday, July 15, 2026· 9 videos

Briefing

Audio player
0:00 / 0:00

OpenAI ships GPT-5.6 Sol

OpenAI integrated Codex directly into ChatGPT, unveiling GPT-5.6 Sol with advanced reasoning and an Ultra mode. The system decomposes complex goals into parallel subtasks handled by multiple agents, accelerating end-to-end development. It also adds deeper computer and browser control, enabling full-stack app workflows from a single interface. More than 6 million weekly users are already engaging with Codex capabilities.

IBM stock plunges 25%

IBM shares fell roughly 25% after resetting expectations for its server and mainframe business. Enterprise spending is shifting toward GPUs, memory, networking, and hyperscale cloud, eroding IBM’s share of AI budgets. Despite strong software margins near 80%, growth is constrained by a services-heavy mix and slowing infrastructure demand. The drop highlights how legacy strengths struggle in the current AI capital cycle.

EU escalates DSA against Meta

European Union regulators intensified action under the Digital Services Act (DSA) targeting Meta’s platform design. Features like infinite scroll, autoplay, and hyper-personalized feeds are flagged as systemic risks that may need to be disabled by default. Penalties could reach 6% of global turnover, potentially exceeding $10 billion. The move signals a shift toward regulating specific addictive mechanics rather than entire platforms.

WhatsApp forced open to rivals

Meta must reopen WhatsApp APIs to competing AI services following EU antitrust pressure. Providers such as OpenAI and Anthropic regain integration access after prior restrictions in 2025. Regulators argue interoperability is essential to fair competition in AI distribution. The decision strengthens Europe’s push to limit platform gatekeeping in messaging ecosystems.

Anthropic launches fear-driven campaign

Anthropic released a stark advertising campaign highlighting AI risks like surveillance, labor disruption, and loss of control. CEO Dario Amodei has warned of a 25% chance of severe outcomes and up to 50% job loss in entry-level white-collar roles within 1–5 years. The campaign drew backlash for its tone, with critics calling it alarmist. It reflects a growing divide over how openly companies should communicate AI risks.

DeepMind urges global AI response

Demis Hassabis of Google DeepMind called for stronger, coordinated action on AI governance. He emphasized the rapid acceleration of capabilities and the need for structured oversight frameworks. Concerns are rising about policymakers’ limited technical understanding of modern AI systems. The comments land amid a temporary slowdown in major product releases but heightened strategic positioning.

Compa maps 285 AI use cases

Compa, a French community project, catalogs over 285 real-world AI applications in a visual “skill tree” interface. The platform shifts focus from tools to use cases, helping users identify needs before selecting solutions. Each entry includes difficulty, automation potential, and links to tools like Midjourney or NotebookLM. It reflects a broader move toward practical AI literacy over tool-centric discovery.

AI detection myths debunked

Common signals like the em dash are not reliable indicators of AI-written text, despite viral claims. Detection tools often perform at only 60–80% accuracy, with OpenAI discontinuing its own after weak results. Models also do not learn directly from individual chats, contrary to popular belief. The findings highlight widespread misunderstanding of how modern AI systems operate.

Videos covered

Previous briefings · AI