Get the newsletter
Microsoft · AI / ML · Filed May 14, 2025 · USPTO published Nov 19, 2026 · This week's release

Microsoft Patents a Second AI That Catches the First One's Made-Up Facts

Microsoft's filing puts one AI to work as a proofreader for another, tagging each shaky sentence with an error type and a score before a third step rewrites it.

An AI watchdog system that reviews text written by an AI helper, catches factual errors, and rewrites the flawed sections before delivery. Drawing from patent filing US 2026/0368153 A1.
An AI watchdog system that reviews text written by an AI helper, catches factual errors, and rewrites the flawed sections before delivery.All 19 drawings ↓
Publication number
US 2026/0368153 A1(US20260368153A1)
Applicant
Microsoft Technology Licensing, LLC
Filed · Published
May 14, 2025 · Nov 19, 2026
Inventors
Robert Tyler Kazuo DESJARDINS, Chungheong GOOI, Chujie HE, Qiang GAN
Status when we published
Approved; patent expected soon
Document
20 claims
US classification
707/691
Examiner
RAAB, CHRISTOPHER J (Art Unit 2156)

What Microsoft's AI fact-checker actually does

Today, a chatbot can answer a question using your documents and still slip in a made-up detail, like a wrong number or a claim your files never made. Developers spend huge effort fine-tuning the instructions they give these bots, but Microsoft's filing says even careful wording can't promise the mistakes disappear.

So the patent adds a second AI as a proofreader. After the first AI writes its answer, the proofreader reads it one sentence at a time and checks each one against your original question and the source material. Shaky sentences get a label, a type of error and a score for how likely they are to be wrong.

Then a rewriter fixes only the flagged sentences, using the same source material, and leaves the rest alone. The system can even show you the first answer right away, then warn you that a corrected version is on the way.

From the filing · CLAIM 1
… a hallucination identifier that receives the model-generated output from the first language model and identifies a first candidate hallucination within the model-generated output using the user query and the associated context, wherein the hallucination identifier comprises a second language model different from the first language model …

Translation: A second AI reviews the first AI's work against the original query and data to spot made-up facts.

How the checker tags and rewrites bad sentences

The system has three roles. A first language model (the chatbot-style AI) answers a user's question using supplied source material, called context. A hallucination identifier, which the claims say must be a second, different language model, then reads that answer passage by passage (roughly sentence by sentence) and compares it with the question and the source material.

For each passage it finds suspicious, it attaches a tag with a category and a score. The filing's categories are contextual consistency (details that clash with the source), numeric accuracy (wrong numbers), and internal coherence (the answer contradicting itself). The checker also writes out its reasoning, which the filing calls an inner monologue, and marks clean passages as clear. In the filing's example, a request to argue that a basketball player deserves MVP, based on a stats dataset, produced an answer where five of seven sentences were flagged.

A rectifier model then gets a prompt (its instructions) built from the flagged passage, its tag, the original question and the source material. It rewrites flagged passages, leaves clean ones intact, and sends the result to the user.

The filing also describes variations:

  • The rewritten answer can go back to the checker for another round, until a quality bar is met.
  • The original answer can be shown first, with an alert that a corrected version is coming, to cut waiting.
  • Without the rewriter, the tags can feed a report used to decide whether to ship an AI update, or to compare how different prompt wordings affect errors.
From the filing · THE ABSTRACT
An LLM-based rectifier is prompted with the hallucination tags and uses the context and the original query to replace the tagged phrases of the original model-generated output.

Translation: A final correction step uses the error tags to rewrite and fix the flawed sentences.

Why a second AI checking the first matters

If you use AI at work to summarize reports or draft documents from your own data, a built-in proofreading step could mean fewer invented figures slipping through. It also takes some pressure off developers, because the filing says careful prompt writing can't promise a fix for every question people might ask. The same tagging can also score how often an AI makes things up, which could help decide whether an update is ready to ship.

The patent office has approved this application and Microsoft has paid the fee to have it issued, so it should become a granted patent soon. The filing doesn't name a product, so where this checking step would show up is unknown.

This is the 52nd Microsoft filing we've tracked since May on AI teams working together, after one on cutting costs by trimming specialist parts and one on handing off to a specialist mid-chat.

Source: US 2026/0368153 A1 at the USPTO (PDF). Full patent text and figures are in the official publication.

Get our take in your Top Stories

Liked this breakdown? Add Patentlyze as a preferred source on Google, and our plain-English take shows up more often in your Top Stories the next time Microsoft patent news breaks.

There are more where this came from

We read every patent application Big Tech publishes and send you the ones worth knowing. Plain English, free, every week.

The drawings · 19 sheets

19 drawing sheets from US 2026/0368153 A1 · click any drawing to enlarge

How a second AI checks answers before delivery, drawing sheet 1 from US 2026/0368153 A1
Sheet 1 How a second AI checks answers before delivery
Inside the two models that spot errors and rewrite answers, drawing sheet 2 from US 2026/0368153 A1
Sheet 2 Inside the two models that spot errors and rewrite answers
The factual source data paired with a person's initial question, drawing sheet 3 from US 2026/0368153 A1
Sheet 3 The factual source data paired with a person's initial question
Draft text from the first AI containing factual errors, drawing sheet 4 from US 2026/0368153 A1
Sheet 4 Draft text from the first AI containing factual errors
Sentence by sentence reasoning used to score each mistake, drawing sheet 5 from US 2026/0368153 A1
Sheet 5 Sentence by sentence reasoning used to score each mistake
Flagged draft text marked with specific error labels and scores, drawing sheet 6 from US 2026/0368153 A1
Sheet 6 Flagged draft text marked with specific error labels and scores
Rewriting the first set of flawed sentences with true facts, drawing sheet 7 from US 2026/0368153 A1
Sheet 7 Rewriting the first set of flawed sentences with true facts
Finishing the corrected rewrite to match the original facts, drawing sheet 8 from US 2026/0368153 A1
Sheet 8 Finishing the corrected rewrite to match the original facts
Show the other 11 sheets
Corrected passages rewritten by the secondary model to fix errors, drawing sheet 9 from US 2026/0368153 A1
Sheet 9 Corrected passages rewritten by the secondary model to fix errors
Final assembled response sent back to the user after verification, drawing sheet 10 from US 2026/0368153 A1
Sheet 10 Final assembled response sent back to the user after verification
Sending corrected answers back through the checker to verify fixes, drawing sheet 11 from US 2026/0368153 A1
Sheet 11 Sending corrected answers back through the checker to verify fixes
Logging errors across multiple answers to evaluate model health, drawing sheet 12 from US 2026/0368153 A1
Sheet 12 Logging errors across multiple answers to evaluate model health
Spotting and tagging inaccurate sentences in an initial response, drawing sheet 13 from US 2026/0368153 A1
Sheet 13 Spotting and tagging inaccurate sentences in an initial response
Alerting users and rewriting faulty text until facts check out, drawing sheet 14 from US 2026/0368153 A1
Sheet 14 Alerting users and rewriting faulty text until facts check out
Deciding to release an AI update after testing, drawing sheet 15 from US 2026/0368153 A1
Sheet 15 Deciding to release an AI update after testing
Tracking recurring errors across separate questions to spot trends, drawing sheet 16 from US 2026/0368153 A1
Sheet 16 Tracking recurring errors across separate questions to spot trends
How the system spots and fixes AI errors before replying, drawing sheet 17 from US 2026/0368153 A1
Sheet 17 How the system spots and fixes AI errors before replying
Tracking AI falsehoods to help developers fix bugs, drawing sheet 18 from US 2026/0368153 A1
Sheet 18 Tracking AI falsehoods to help developers fix bugs
Computer parts that power the error-checking service, drawing sheet 19 from US 2026/0368153 A1
Sheet 19 Computer parts that power the error-checking service
Patent filing page

Watchlists this filing belongs to

All watchlists →
Reader comments

No comments yet

Start the discussion

Real name or a handle, either is fine. Comments are read by a person before they appear, so allow a little time. Keep it about the filing.