🔍 Read the full analysis: How AI Found A Hidden Message In The Stars on ThorstenMeyerAI.com
Get business pricing on tech for your team
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
A Gizmodo headline says an OpenAI bot found a secret message that an Anthropic AI hid in a star field. The article body, test details and independent verification were unavailable, so the headline alone does not establish how the event happened or what it shows about either system.
A Gizmodo headline says an OpenAI bot found a message that an Anthropic AI had hidden in a star field, describing an exchange between systems associated with two AI companies, as recounted in the original analysis. The article body and supporting evidence were unavailable in the information reviewed, so the reported event’s method, circumstances and result cannot be independently established.
The headline, “Anthropic’s AI Hid a Secret Message in the Stars. OpenAI’s Bot Found It,” is the only account available. It names Anthropic and OpenAI, but does not identify particular systems or model versions. It also does not explain what “in the stars” means: the message might have appeared in an image, data, or another kind of task, but the details are not specified.
No test procedure, transcript, image, evaluation or statements from the companies or participants were provided. The available information does not say how the message was encoded, what the OpenAI system was asked to do, or what evidence supports the claim that it found the message. Nor does it establish whether the exchange was a controlled experiment, a demonstration, a competition or an informal interaction.
The headline presents a clear outcome, but it does not reveal how success was judged or whether the result was repeated. There are no details about number of attempts, comparison conditions or independent replication. The confirmed development is limited: Gizmodo published a headline making the claim. The underlying event and its technical details remain unverified in the material available.
What a Verified Test Could Show
If the headline describes a real, controlled test, it could offer an example of one AI system detecting information another deliberately concealed. That would be relevant to people studying hidden-pattern detection and model behavior, as well as to those evaluating how systems handle information placed in visual or other inputs.
But an individual demonstration would not, by itself, show that one company’s models generally outperform another’s, or that either system can reliably hide or detect messages across different tasks. Without the test design and evaluation criteria, readers cannot tell whether the message was difficult to find, whether clues made it conspicuous, or whether the result depended on a narrowly tailored prompt. The headline is not evidence of broad capability on its own.
A documented test could still be informative even if it were a single example. Its value would depend on whether observers could see how the message was hidden, what the detection system received, and how success was determined. Those details would help distinguish a meaningful result from a simple or selectively presented demonstration.
As an affiliate, we earn on qualifying purchases.
The Headline Lacks Test Details
The reported claim centers on a cross-company interaction: an Anthropic AI is said to have hidden a message, and an OpenAI bot is said to have found it. The available account does not provide a publication date for the headline or the date when the alleged test took place. It also gives no model names or versions, making it impossible to connect the claim to particular systems or assess whether it applies to any current model.
The phrase “in the stars” is not explained. It could describe a visual star field or another representation, but the available details do not support choosing among those possibilities. There is no description of prompts, task instructions, the hidden message itself, or the criteria used to decide whether it had been found. These omissions prevent readers from reconstructing the task or judging its difficulty.
No direct comments from Anthropic, OpenAI, researchers or other participants accompany the information available. There is also no indication that the reported outcome was checked by an independent party. Neither company’s confirmation nor a repeatable demonstration is established by the headline alone.
“Anthropic’s AI Hid a Secret Message in the Stars. OpenAI’s Bot Found It.”
— Gizmodo headline
computer vision and image recognition book
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Evidence Is Still Missing
It remains unclear which systems were involved, how the message was hidden, and what the OpenAI bot was given to detect it. The available account does not say whether “stars” refers to image pixels, symbolic data or a narrative task. It also does not explain how success was scored, whether the result was repeated, or whether a human could readily spot the message.
There is no supplied evidence that either company confirmed the event, and no information about the test’s date, number of trials or baseline for comparison. Without those details, it is not possible to assess whether the outcome was robust, a one-off example or an account that needs further verification. The claim should be treated as unverified on the information available, rather than as proof of a general capability.
As an affiliate, we earn on qualifying purchases.
What Verification Would Require
A fuller report could clarify the claim by naming the systems and versions, describing the task and publishing the prompts or relevant materials. Readers would also need to know how many attempts were made, what counted as a successful detection, and whether the procedure was repeated under comparable conditions.
Statements from Anthropic, OpenAI or the people who conducted the test could help establish what happened, though a clear and reproducible demonstration would allow the result to be examined more directly. No follow-up report, response or additional testing is identified in the available information. Until those details appear, the headline remains the limit of what can be confirmed.
prompt engineering guide for large language models
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Questions
What does the Gizmodo headline claim?
It says an OpenAI bot found a message that an Anthropic AI had hidden in a star field. The supporting article details were not available in the account reviewed.
Which AI systems were involved?
The headline names Anthropic and OpenAI, but does not identify specific models or versions.
How was the message hidden?
The method is not specified. The available information does not say whether the star field was an image, data or another kind of task.
Has the result been independently verified?
No verification details were provided. The headline alone does not establish whether the companies confirmed the event or whether the result was repeated.
Primary source: Anthropic · via ThorstenMeyerAI.com
Halloween Picks
halloween
As an affiliate, we earn on qualifying purchases.
