Myths
Myths about neuralese
Neuralese attracts myths from both directions. The alarmed ones say AI is plotting in a secret tongue. The comfortable ones say that if a model writes its reasoning in English, we can read its mind.
Here’s each one we track, with where the evidence stands and the sources behind it.
- Myths tracked
- 14
01
About the word
- Wrong
#The word “neuralese” was made up in 2025 for the AI 2027 forecast.
It goes back at least to 2017, when Jacob Andreas, Anca Dragan and Dan Klein used it in a paper called “Translating Neuralese” for the messages AI agents had learned to send each other. AI 2027 brought it to a wide audience in 2025, in its newer sense: a model reasoning in its own numbers.
- Wrong
#Neuralese is a language, with its own words and grammar.
It’s the stream of numbers a model works with inside, used in place of words. There’s no vocabulary or grammar to crack. People who try to read it translate or probe the numbers instead, the way the 2017 paper matched each message with the English phrase that left a listener believing the same thing.
- Not established
#“Neuralese” is OpenAI’s term for how Astra works.
The word appears nowhere in the system card or in Pachocki’s essay. It does appear in Microsoft’s draft code of conduct, which isn’t about Astra.
- Wrong
#Microsoft has banned its AI from using neuralese.
Microsoft AI proposed that its models “do not communicate in neuralese or any form beyond simple human understanding.” It’s a draft, open for public comment and not in force, and Microsoft says it isn’t training models on it yet.
02
About how it works
- Wrong
#Thinking in numbers makes a model 1,000 times smarter.
The figure comes from AI 2027, which put the gap between a written word and a model’s internal state at “over 1,000 times more information.” That’s about how much one step can carry, not how well a model reasons. COCONUT, the best-known method, ran on a small GPT-2 model and scored 34% on grade-school math, against 43% when the same model wrote its reasoning out.
- Wrong
#Continuous thoughts and recurrent depth are the same idea.
They change different things. Continuous thoughts, as in COCONUT, replace written steps: the model feeds its internal state back in as its next thought instead of writing a word. Recurrent depth runs a block of the model’s layers in a loop, so each step gets more computation. A model could use either, both or neither. How the two differ.
- Open question
#Reasoning without words already works better than writing it out.
It depends on the task. COCONUT searched more widely on planning puzzles but trailed written reasoning on math. A recurrent-depth model with 3.5 billion parameters raised its scores by looping more, up to the computation of a 50-billion-parameter model. Neither has been shown at the scale of today’s frontier models.
- Wrong
#Nobody is building AI that reasons without words.
Researchers have published working methods for years. COCONUT feeds a model’s internal state back in as its next thought instead of writing a word, and recurrent-depth models run their own layers in a loop to think longer before they answer. The Information reported that GPT-6 Astra uses recurrent depth. OpenAI hasn’t confirmed it.
- Wrong
#As long as a model reasons in plain English, we can see what it’s thinking.
Written reasoning leaves things out. In Anthropic’s tests, when a slipped-in hint changed a model’s answer, Claude 3.7 Sonnet mentioned the hint only 25% of the time and DeepSeek R1 39%. In OpenAI’s experiments, punishing a model for suspicious reasoning taught it to hide its intent while it kept cheating.
03
About GPT-6 Astra
- Wrong
#Astra has stopped thinking in words.
Astra still writes a chain of thought, and most of OpenAI’s monitorability testing is about reading it. What changed is how much that chain of thought gives away.
- Not established
#Astra reasons in COCONUT-style continuous thoughts.
Nothing public says so. The reporting is about recurrent depth, which is a different technique. How the two differ.
- Not established
#Recurrent depth caused the drop in monitorability.
OpenAI says it is “quite confident” that the change in controllability is not “differentially due to any architectural changes,” and that the change tracks Astra’s gains at working without reasoning. In his 6 September essay, Pachocki blamed “a combination of factors”: more complex, tool-heavy tasks, models getting better at manipulating their own reasoning, and models becoming “much smarter even without using verbalized reasoning at all.” Architecture isn’t on his list.
These come from the Astra ledger, which tracks every claim about the model and what would change each status.
04
About secret languages
- Not established
#AI models are already talking to each other in a secret language.
We know of no evidence that deployed models pass hidden messages to each other on their own. Researchers do test for the nearest real thing, a model tucking hidden reasoning into ordinary-looking text. OpenAI’s evaluations of GPT-6 Astra “indicate Astra is not able to perform steganographic CoT reasoning.”
- Wrong
#When AI drifts into its own shorthand, it’s hiding something.
Drift like the Facebook bots’ is a known side effect of training. When nothing rewards a model for staying readable, it takes whatever shortcut pays, the way traders or surgeons build their own jargon. Dhruv Batra called it “a well-established sub-field of AI, with publications dating back decades.”
And the viral ones, each told in full on Does AI have a secret language?
05
Sources
Every myth on this page traces to one of these. All were checked on 24 September 2026, against the original text wherever we could read it. The viral stories have their sources on their own page.
- Peer-reviewed paper
Jacob Andreas, Anca Dragan, Dan Klein. Translating Neuralese. ACL 2017.
Where the word comes from, and how the messages were translated.
- Forecast
Daniel Kokotajlo, Scott Alexander, Thomas Larsen, Eli Lifland, Romeo Dean. AI 2027. April 2025.
The word’s wide audience, and the “1,000 times” figure.
- Draft policy
Microsoft AI. Humanist AI Code of Conduct for MAI Models. Draft for public consultation, 14 Sep 2026.
Microsoft’s draft rule.
- Peer-reviewed paper
Shibo Hao et al. Training Large Language Models to Reason in a Continuous Latent Space. COLM 2025.
Continuous thoughts, and how they did on math and planning.
- Peer-reviewed paper
Jonas Geiping et al. Scaling up Test-Time Compute with Latent Reasoning: A Recurrent Depth Approach. NeurIPS 2025.
Recurrent depth.
- Lab publication
Anthropic. Reasoning models don’t always say what they think. April 2025.
How often models mention the hints they used.
- Preprint
Bowen Baker et al. (OpenAI). Monitoring Reasoning Models for Misbehavior and the Risks of Promoting Obfuscation. March 2025.
Models learning to hide their intent.
- Lab publication
OpenAI. GPT-6 Astra System Card. 3 Sep 2026, updated 9 and 22 Sep 2026.
What OpenAI says about Astra.
- Lab publication
Jakub Pachocki (OpenAI). An Alien Mind. 6 Sep 2026.
Pachocki on Astra.
- Journalism
Amir Efrati, Stephanie Palazzolo, Rocket Drew. OpenAI Technique in ‘Astra’ Model Sparks Security Concerns. The Information, 1 Sep 2026.
The report that Astra uses recurrent depth. Paywalled. We haven’t read the full text, and the details here match how other outlets summarized it.
- Journalism
Mark Wilson. AI Is Inventing Languages Humans Can’t Understand. Should We Stop It? Fast Company, 14 Jul 2017.
Why bots drift into shorthand.
- Journalism
USA Today. Fact check: Facebook didn’t pull the plug on two chatbots because they created a language. USA Today, 28 Jul 2021.
Dhruv Batra on drift as a known effect. Linked to the Tech Xplore reprint of the USA Today article.