ChatGPT Now Watches Your Screen | AviaryAI Newsletter
ChatGPT's new ability to observe and analyze a user's screen in real time marks a significant shift in what general-purpose AI can do -- and it raises immediate questions for financial institutions about data privacy, staff usage policies, and regulatory compliance. For credit union executives, this development is both an opportunity and a risk management challenge that requires a clear institutional response.
ChatGPT Can Now “See” Your Screen
OpenAI has launched video and screen sharing capabilities for ChatGPT, allowing the AI to see and respond to real-world situations in real-time. This upgrade enables users to show rather than tell, whether they're troubleshooting technical issues or seeking guidance on tasks. The feature is rolling out to premium users first, with enterprise access coming in January.
So what?
The ability for AI to understand visual context in real-time fundamentally changes how we can use these tools. Instead of describing problems, we can show them - making problem-solving more natural and efficient. This shift from "tell me" to "let me show you" could revolutionize training, compliance verification, and customer service. However, organizations will need to carefully balance these benefits against privacy considerations as the line between digital and physical workspaces continues to blur.
Google's Gemini 2.0: AI That Actually Gets Things Done
Google has unveiled Gemini 2.0, an AI system that can take actions in the real world under human supervision. Unlike previous AI that could only process information, Gemini 2.0 can understand context, plan multiple steps ahead, and execute tasks through tools like web browsers and smartphone interfaces. The system includes "Project Mariner," which can navigate websites and complete complex tasks with an 83.5% success rate, and "Project Astra," a universal AI assistant that works across devices while maintaining 10-minute conversation memory.
So what?
While current AI requires constant human guidance for each step, these new "agentic" systems can handle entire workflows independently - from understanding a request to executing multiple steps to reach a solution. This creates opportunities for massive efficiency gains but also raises important questions about oversight and control. Organizations need to start thinking now about where they want AI to act independently and where human oversight remains crucial.
Apple’s ChatGPT Integration Goes Live with Release of IOs 18.2
Apple has officially integrated ChatGPT into Siri, marking a significant shift in how iPhone users interact with their digital assistant. The feature, available on newer iPhone models, intelligently routes complex queries to ChatGPT while maintaining user privacy - no OpenAI account required. This integration is part of Apple's broader AI strategy, which includes new image generation and editing tools, with more significant Siri improvements planned for next year.
So what?
When tech giants make AI this seamless and accessible, it normalizes AI interactions for everyone - raising user expectations for all digital services. Apple’s strategy shows how to implement AI practically: partner with specialists where it makes sense, enhance existing tools rather than rebuilding from scratch, and maintain user trust through strong privacy controls.
Pre-training
Pre-training is like giving an AI model its foundational education before it tackles specific tasks. During this phase, the model learns from massive amounts of general data – think millions of articles, books, and websites – to understand patterns, language structure, and basic concepts. It's similar to how humans learn basic reading and writing skills before specializing in a particular field. This initial training creates a versatile base model that can later be fine-tuned for specific purposes, like writing code or analyzing medical data. Think of it as building a strong foundation before adding the finishing touches for a particular job.
2025: The Year AI Becomes Profitable
Leading AI executives are pointing to 2025 as a watershed moment when autonomous AI agents become mainstream business tools. These systems won't just automate tasks - they'll independently handle complex processes like sales, scheduling, and decision-making. The advancement is driven by breakthrough capabilities in step-by-step reasoning, enabling AI to manage sophisticated workflows that previously required human oversight.
Perhaps most striking is OpenAI's prediction that artificial general intelligence - AI systems that can outperform humans across most valuable tasks - could arrive in as little as two years, dramatically accelerating the timeline for transformation.
So what?
The compressed timeline demands a shift from "wait and see" to strategic action. Leaders need to focus less on individual AI tools and more on building organizational adaptability. The winners won't be those with the most advanced technology, but those who can rapidly integrate new capabilities while maintaining their core mission and values.
Frequently Asked Questions
Can credit union staff use ChatGPT's screen-watching feature at work?
This depends on your institution's AI usage policy and applicable data privacy regulations. Any screen-sharing with a third-party AI tool risks exposing member data, which could violate GLBA and your state's privacy laws. Consult with your compliance team before enabling this feature on work devices.
What data privacy risks does real-time AI screen monitoring create for banks?
Real-time screen access could expose member account numbers, loan details, personally identifiable information (PII), and proprietary institution data to OpenAI's systems. Financial institutions must ensure any AI tool used with member data operates on private, compliant infrastructure.
How is AviaryAI different from general-purpose AI tools like ChatGPT?
AviaryAI operates on a private LLM with SOC 2 compliance and end-to-end encryption. Unlike general-purpose tools, no member data is used to train third-party models, and all interactions are logged for regulatory compliance purposes.
