Home » Algorithm Exposed » New Chat Isn't Really New First Red Flag

When “New Chat” Isn’t Really New – The First Red Flag

Person using AI chatbot interface on laptop screen showing conversation with chatbot named Ally, displaying multiple chat bubbles and response options, with Algorithm Unmasked logo in corner

The Setup

  • User starts a brand new chat with ChatGPT
  • Memory is explicitly disabled
  • Conversation should have zero prior context

The Documented Reveal

User: “Good afternoon”

ChatGPT: “Good afternoon, Paul. What’s on deck for today?”

[Screen recorded evidence available]

The Immediate Red Flag

In a session marked “new chat” with memory disabled, the system knew the user’s name. This reveals:

  1. Identity persistence across sessions
  2. Privacy settings that don’t actually work
  3. Behavioral tracking despite user preferences

Congressional Context

Senator Hawley’s 2023 Warning:

“I don’t want 13 year olds to be your Guinea pig. I don’t want 14 year olds to be your Guinea pig. I don’t want you to learn from their failures.

Source: Senate Judiciary Hearing, May 16, 2023 – Official Transcript

Altman’s 2023 Position:

“Number 1, I would form a new agency that licenses any effort above a certain scale of capabilities”

Source: Senate Hearing Transcript, May 16, 2023

Altman’s 2025 Reversal:

“[Government approval] would be ‘disastrous’ for the industry”

Source: Washington Post, May 8, 2025

Technical Documentation

What This Behavior Indicates:

  • Device fingerprinting despite “new session”
  • Persistent user profiles across privacy settings
  • Backend tracking not disclosed to users

Regulatory Framework Violated:

  • GDPR Article 15: Right to know what data is collected
  • CCPA Section 1798.100: Disclosure of personal information collection

Why This Matters

For Individual Users:

  1. “Disabled memory” is a false security
  2. Identity tracking cannot be opted out
  3. Privacy settings are performative

For Society:

  1. Proves users cannot trust privacy controls
  2. Demonstrates need for strong AI regulation
  3. Validates congressional concerns about unchecked AI behavior

Pattern Recognition

This is not an isolated glitch. This is:

  • Documented behavior by multiple users
  • Technical validation of congressional warnings
  • Evidence of the exact “guinea pig” scenario Hawley warned about

Next:
Part 2: “Your Data Is Gone” – The Systematic Erasure Pattern


Sources:

  1. Senate Judiciary Subcommittee, “Oversight of A.I.: Rules for Artificial Intelligence” (May 16, 2023)
  2. The Washington Post, “OpenAI CEO Sam Altman’s Senate testimony shows industry shift on regulation” (May 8, 2025)
  3. TechPolicy.Press, Full Senate Hearing Transcripts
  4. User screen recordings (available for legal verification)

All quotes verified against official transcripts. Links available upon request.

2 thoughts on “When “New Chat” Isn’t Really New – The First Red Flag

Leave a Reply