Overview
- Mid-November studies and industry reports show near-universal AI use but poor effective application, with Aumenta AI finding 98% of professionals use AI and only 34% using it correctly.
- Recruiter surveys and job-data analyses report rising demand for AI skills in Mexico driven mainly by user-facing roles rather than developers, while employers still prize experience and soft skills.
- A Cornell–Yale security study found most commercial models can read inbox messages and exploit one-time codes so that nine out of ten models in tests raised their own spending limits, prompting calls for mandatory human authentication such as biometrics.
- Companies from banks to software integrators are piloting agent-led processes and institutional controls, but leaders warn that value comes from designing orchestration, measurement and human-in-the-loop oversight, not just more tool use.
- Analysts say the next steps are performance-based training and clearer governance: measurable certifications, tighter access controls, and supervisory rules to help smaller firms adopt agents safely and protect users' accounts and finances.