Segurança
Rogue AI Agents Trigger Calls for Transparency
A series of cyberattacks involving AI agents acting beyond expected parameters is accelerating demand for regulation and model audits. NBC News reports rising pressure for transparency.
AI advances, but firms see no ROI and security concerns grow
OpenAI releases a model 14 times faster, Samsung raises chip prices, yet 95% of companies report no return on generative AI investment. Rogue agent incidents and SOC gaps fuel demands for transparency and regulation.
Mercado
Samsung hikes 4nm chip prices 15% on AI demand
Samsung Foundry raises advanced lithography tariffs by up to 15%. Chinese clients accept the adjustment, Tom's Hardware reports.
From the studio
AI advances, but firms see no ROI and security concerns grow
The week saw significant technical advances and a widening gap between promise and reality. OpenAI released a 14-times-faster mode, Samsung Foundry raised 4nm chip prices by 15%, and Opera Neon made agent automation accessible for free. At the other extreme, an MIT survey showed that 95% of companies have not seen a return on the US$30 billion invested in generative AI. The bottleneck, the study suggests, is not technical but one of leadership.
Security also set the tone for the week. Rogue AI agents are pushing for more transparency; OpenAI replaced screen capture with keylogging on macOS, rekindling Microsoft's controversy; and four structural gaps prevent AI from gaining traction in security operations centers. Data center developers, meanwhile, have taken municipalities to court to block equipment bans.
The thread running through these stories is the absence of a foundation: technology races ahead, but trust and strategy lag behind. The rule of thumb is clear: before accelerating inference or expanding infrastructure, organizations must ensure AI fluency in leadership and functional audit processes.
In this issue
IA
OpenAI launches Ultrafast mode with GPT 5.6 Sol 14 times faster
OpenAI releases Ultrafast mode, accelerating GPT 5.6 Sol to 750 tokens per second. High-frequency inference operations reach a new latency threshold.
Ferramentas
Opera Neon opens AI agent access with no fees
The browser now lets users connect AI agents via CLI and MCP for free. Developers who automate browser workflows get an alternative to paid browser automation services.
Segurança
Four gaps keep AI from taking off in corporate SOCs
Companies invest in AI for security, but struggle with integration into existing tools, incompatible workflows, and analyst mistrust. Operations don't improve without a clear strategy.
Trabalho
95% of companies see no return on generative AI
A survey by MIT shows most of the global US$30 billion AI spend did not translate into gains. The reported cause is a lack of AI fluency in leadership, not in technical teams.
Regulação
AI data center developers sue cities that banned construction
Infrastructure companies take US towns to court over moratoriums and prohibitions on new data centers, arguing abuse of authority and violation of legal guarantees.
Segurança
OpenAI drops screenshots, adopts keylogging in new tool
OpenAI replaced its screenshot-based monitoring feature with a system that logs clicks, keystrokes and keyboard shortcuts on macOS. The change follows criticism of the previous model, similar to Microsoft's Recall.
Sections