Pull to refresh
Logo
Daily Brief
Following
Why Ranks Sign Up
OpenAI gives ChatGPT self-updating long-term memory

OpenAI gives ChatGPT self-updating long-term memory

New Capabilities

The 'Dreaming V3' system synthesizes a user's history across years of chats, then ages out what's stale

Today: Dreaming V3 launches

Overview

For most of ChatGPT's life, you had to tell it to remember things, one fact at a time. On June 5, 2026, OpenAI changed that. Its new 'Dreaming V3' system reads across years of your past chats and writes its own running profile of you, no prompting required.

The system rolled out first to paying Plus and Pro users in the United States. OpenAI says it cut the computing cost of running the feature by about five times, which makes a free-tier version possible within weeks. That turns a chatbot you query into an assistant that carries context forward, and it puts a standing record of your preferences inside a company's servers.

Why it matters

ChatGPT now builds a quiet, self-updating dossier on each user, raising what it can do for you and how much it knows about you.

Questions about this story

No questions yet — be the first to ask.

Key Indicators

~5x
Compute cut
OpenAI says optimizations reduced the computing power to serve the feature to free users by roughly five times.
30 days
Deleted-memory retention
Logs of deleted saved memories may be kept up to 30 days for safety and debugging.
82.8%
Factual recall rate
Reported success on factual recall tasks, up from 41.5% in earlier testing.
Aug 2026
EU AI Act duties begin
New compliance obligations under the EU AI Act start in August 2026.

Voices

Curated perspectives — historical figures and your fellow readers.

Ever wondered what historical figures would say about today's headlines?

Sign up to generate historical perspectives on this story.

Play

Exploring all sides of a story is often best achieved with Play.

Log in to play. Track your picks, climb the leaderboards. Log in Sign Up
Predict 3 ways this could play out. Contrarian picks score more — points lock when the scenario resolves. Log in to play
Higher or Lower Two numbers from this story. Guess which is bigger. 5 rounds to set a streak. Log in to play
Connections Sixteen names from the news. Find the four hidden groups of four. Log in to play

People Involved

Organizations Involved

Timeline

April 2025 June 2026

4 events Latest: Today
Tap a bar to jump to that date
  1. Dreaming V3 launches

    Today Product

    OpenAI releases a rebuilt memory system to US Plus and Pro users. It auto-synthesizes context across years of chats and lets users view, edit, or delete entries.

  2. Google expands Gemini memory

    Competition

    Google rolls out 'Personal Intelligence' for Gemini in the US, letting it draw on Gmail, Photos, and other Google data.

  3. Anthropic gives Claude memory

    Competition

    Anthropic begins letting Claude remember prior interactions for paid users, sharpening the race on personalization.

  4. OpenAI introduces 'dreaming'

    Product

    OpenAI adds a background process that learns from past conversations and synthesizes memory, the foundation for later versions.

Historical Context

2 moments from history that rhyme with this story — and how they unfolded.

April 2004

Gmail launches with content scanning (2004)

Google launched Gmail with then-huge free storage, paid for by scanning email content to target ads. Privacy groups objected that machines reading private mail crossed a line, even if no human read it.

Then

Gmail drew complaints and regulatory questions but grew fast on its free storage and speed.

Now

Google stopped scanning Gmail for ad targeting in 2017, after the practice shaped years of privacy debate.

Why this matters now

Like Gmail, Dreaming V3 trades deeper machine access to your data for a better free product. The same question returns: does processing your private content for utility count as a privacy cost?

September 2006

Facebook News Feed backlash (2006)

Facebook turned scattered profile updates into a single feed that broadcast everyone's activity. Users revolted, saying the same information now felt exposed because it was aggregated and surfaced automatically.

Then

Facebook kept the feature but added privacy controls within days after mass protest.

Now

News Feed became the core of the product and the model for social media.

Why this matters now

Synthesis is the issue in both cases. Facts users shared freely feel different once a system stitches them together and acts on them without being asked, which is exactly what 'context bleed' complaints describe.

Sources

(6)