Skip to content
IIAIT

AI Today · 20 November 2026

Google releases Gemini 4 Neon, and more US banks make customers liable for third-party AI agents

A faster, lower-priced Gemini model reaches paid plans and the API, more large US banks shift the risk of outside AI agents onto customers, and California sets out how it wants frontier AI to be overseen.

  1. 01

    Google releases Gemini 4 Neon, a mid-tier model priced at $3 per million input tokens

    Google released Gemini 4 Neon on November 18, 2026, placing it between Gemini 4 Carbon and Gemini 4 Hydrogen and aiming it at coding, knowledge work and science. It is available in the Gemini API through Google AI Studio, in Gemini for business, and in the Gemini app for Google AI Pro and Ultra subscribers. API pricing is $3 per million input tokens, $0.15 for cached input and $15 per million output tokens, with a 1 million token context window. The Decoder reports it runs about twice as fast as Carbon and ranks third on the Artificial Analysis Intelligence Index, behind models from OpenAI and Anthropic that Google left out of its own comparison table.

    What it means for you

    If you pay for Google AI Pro or Ultra or use Gemini for business, compare Neon with the model you use now on a few of your real tasks before switching. Benchmark rankings come from Google and third-party leaderboards, so check output quality and cost per task yourself.

  2. 02

    Citi, PNC and Citizens tell customers they are responsible for what third-party AI agents do in their accounts

    TechCrunch reported on November 18, 2026 that Citi updated its Online User Agreement so that customers who give third-party AI tools access to their accounts are responsible for activity through those tools, including if the tool is hacked or leaks credentials. PNC changed its terms the same day and Citizens sent customers a similar warning. JPMorgan Chase and Bank of America announced comparable policies in October. Under Citi's terms, AI agents the bank provides itself can be used without that added liability. American Banker reported earlier that Wells Fargo's updated terms, effective November 24, keep customers responsible for an AI agent's activity until they give notice that its access was revoked.

    What it means for you

    Before connecting an AI assistant to a personal or business bank account, read your bank's online terms and prefer read-only connections. If you have already shared banking credentials with an AI tool, consider revoking that access in writing and changing your password.

  3. 03

    California recommends mandatory AI safety standards, independent testing and state power to order emergency shutdowns

    On November 18, 2026, Governor Newsom's office released recommendations prepared under his September executive order on AI safety. They call for state-set baseline safety standards for AI companies, independent testing during development, expanded incident reporting, verified procedures for shutting down dangerous models, state authority to order an emergency shutdown, and power to cap recursive self-improvement, where an AI system builds its own successor. Frontier developers would have to let independent evaluators accountable to the state assess their models. These are recommendations to the Legislature, not law, and Los Angeles magazine notes that the incoming governor and lawmakers will decide whether they advance.

    What it means for you

    Nothing changes yet for businesses that use AI tools, since the measures would apply mainly to developers and need legislation first. If you rely on frontier models in your work, follow the bills next year, because shutdown and reporting rules could affect how vendors operate and disclose incidents.