NASA UAP Report 2023: No ET Evidence, Spergel Panel Urges AI Tools

You keep seeing “UFO disclosure” and “UAP news” headlines that read like a countdown to alien confirmation. Then you run into NASA’s 2023 report and realize it doesn’t play that game.

The real frustration is simple: you want a clean answer. Is it a government UFO cover-up, a stack of mundane misreads, or something genuinely unknown? NASA’s message lands in the most annoying middle ground for anyone chasing certainty: a lot of cases can’t be settled because the underlying data and eyewitness accounts aren’t strong enough to support a definitive call.

That’s the tradeoff the internet rarely admits. Certainty requires higher-quality measurements, consistent reporting, and more transparency around what was captured and how, and that process moves at the speed of methodical verification, not viral clips. The upside is that this framing actually helps you think clearly about “no ET evidence” without jumping to “nothing happened” or “cover-up confirmed.” It’s a quality-bar problem, not a verdict.

NASA made that posture explicit on September 14, 2023, when it released its independent study team’s report alongside a NASA media briefing at NASA HQ in Washington at 10 a.m. EDT. The headline conclusion was blunt: the team “did not find any evidence that UAP have an extraterrestrial origin” (NASA UAP Independent Study Team report (PDF); NASA press release and briefing). That line disappoints people expecting non-human intelligence to be confirmed, but it also does something more useful. It tells you NASA is treating this as a data problem first, because without better inputs, “unidentified” stays stuck in limbo.

Following the report, the agency created a Director of UAP Research role and appointed Mark McInerney, with a mission that includes coordinating NASA’s work to understand the causes of UAP sightings (NASA announcement). That’s a concrete signal the agency sees this as a real research and coordination effort, not a one-off headline.

You’ll leave this with a cleaner way to interpret the report, separate “unidentified” from “unexplainable,” and know exactly what kind of evidence would actually change the story next.

Inside the Spergel UAP Panel

To understand why NASA lands where it does, it helps to look at what this panel was actually built to do. If you treat this like a scientific scoping document, it suddenly makes more sense.

Read NASA’s UAP report as a methodology and data intervention, not a verdict on “are aliens real.” The whole point is to tighten up how weird reports get collected, described, and analyzed so future conclusions are driven by usable evidence instead of vibes.

NASA commissioned an independent study team in June of the previous year to examine UAP and produce findings and recommendations to inform NASA about available data. That “inform NASA about available data” phrasing tells you the problem they were trying to solve: the data is messy, scattered across systems, and often not collected in a way that supports scientific follow-up.

Public interest and political pressure are the backdrop, but the technical bottleneck is basic: NASA itself has pointed to the need for a data-driven roadmap, and reporting on NASA’s position notes that existing data and eyewitness accounts are often insufficient for conclusive determinations in many cases. That’s why the report reads less like a case file and more like an engineering handoff: define what “good data” should look like, then build toward it.

The report’s bottom line stays consistent with that mission: it doesn’t treat limited, uneven observations as a basis for big claims, and it focuses on what would make future evaluations more defensible.

NASA selected Dr. David Spergel to chair the UAP Independent Study Team, and NASA materials identify him as president of the Simons Foundation. That choice is a credibility signal: you’re looking at a chaired panel structured like a science-facing review, not an ad hoc group chasing headlines.

The roster includes names like Anamaria Berea and Federica Bianco, which gives you a feel for the posture: cross-disciplinary expertise aimed at measurement, analysis, and astronomy-adjacent data problems, not a “we’re going to kick down doors and solve every incident” mandate.

That last point matters because people often assume “NASA study” means “NASA investigation.” This effort wasn’t set up with investigative authority or a law enforcement vibe. It’s a civilian science framing: look at what data exists, what data could exist, and how NASA can responsibly contribute tools and standards using unclassified or publicly usable methods.

NASA’s definition of Unidentified Anomalous Phenomena (UAP) is intentionally broad: “phenomena or observations of events in the air, sea, space, and land that cannot be identified as aircraft or known natural phenomena.” In practice, that scope choice is a constraint as much as it is an invitation, because it pulls in everything from sensor oddities to rare atmospheric effects, and the report has to operate inside that uncertainty.

It also explains why NASA treats “UAP” as an observation category, not a conclusion. The report is built to improve how observations are captured and compared, not to smuggle in an assumption about what’s behind them.

Then there’s the other acronym you’ll see everywhere: “The Department of Defense’s All-domain Anomaly Resolution Office (AARO) is a military office tasked with investigating UAP.” That one sentence clarifies why NASA’s work looks so different. AARO is oriented around investigation inside a defense context; NASA’s panel is oriented around science-grade data practices that can survive public scrutiny and be used by civilian researchers.

Here’s the mental model that keeps you from misreading the report: it can realistically answer what data is available, what data is missing, and what standards would let future UAP reports be evaluated consistently. It cannot adjudicate every sighting, substitute for a military investigative office, or turn limited, uneven observations into a final ruling about what every incident “really was.”

No ET Evidence And Why

That scope-standards and data quality instead of case-by-case adjudication-sets up the report’s most misunderstood line. When the inputs are messy, the most honest output is usually restraint.

The report doesn’t say “nothing exists”, it says “we can’t conclude much from messy inputs.” In practice, “unidentified” usually means “not enough reliable information,” not “non-human intelligence,” and the panel’s “no ET evidence” outcome is exactly what you’d expect when the data arrives incomplete or non-repeatable.

NASA’s plain-language explanation is also the most practical one: a lot of UAP investigations stall because the underlying observations are hard to validate, hard to reproduce, and missing the context needed to run a serious analysis.

1. Poor sensor calibration. Sensor calibration is the process of verifying and adjusting a sensor so its measurements match known standards; if you can’t show when a camera, radar, or IR system was last calibrated, you can’t tell whether an odd “speed” reading reflects the object or the instrument. That’s how a clip becomes a debate about aliens instead of a straightforward equipment question.
2. Single-sensor observations (lack of multiple measurements). One instrument, one viewpoint, one moment in time can look dramatic while still being ambiguous. A lone camera can’t directly give you range, and without range you can’t reliably calculate size or speed. That’s why NASA flags the lack of multiple measurements as a core reason cases stay unresolved.
3. Missing sensor metadata. Metadata is the descriptive information that travels with a file (time stamps, location, sensor settings, platform orientation); when metadata is absent or stripped, you can’t reconstruct the observation well enough to test explanations. A video without time, lens parameters, or platform heading forces analysts to guess, and guesses don’t close cases.

The friction is that calibration details often live outside the video itself: in maintenance logs, engineering documentation, or system health reports. If those aren’t attached to the case file, investigators are left reverse-engineering the sensor from a compressed snippet. The actionable takeaway: a sighting isn’t just “what the sensor saw,” it’s “what the sensor is proven capable of measuring.”

The catch: people assume HD video equals certainty. It doesn’t. A clean image can still be missing the one variable that matters most: distance. The fix is simple in concept even if hard in execution: corroborate with a second sensor type (for example, radar plus imagery) or a second geometry (another camera angle).

The real-world nuance is that metadata often gets lost during perfectly normal sharing: screen recordings, social uploads, re-encodes, edits, and reposts. If the goal is “solve it,” the original file matters more than the most-shared copy.

NASA explicitly flags parallax effects from changing sensor geometry and irregular time sampling as sources of undesirable effects that can drive misinterpretation. Here’s the intuitive version: if the sensor is moving (aircraft, satellite, vehicle) and the object is far away, tiny changes in viewpoint can make an object appear to dart, hover, or change shape even when it’s behaving normally.

Irregular sampling adds another trap. If frames or measurements arrive at uneven intervals, the “in-between” motion gets filled in by your brain or by naive calculations, which can inflate perceived acceleration or create weird trajectories. Atmospherics and sensor artifacts can layer on top of that, but you usually don’t need exotic physics to explain why a short clip looks impossible.

Accurate time synchronization is a known and persistent issue in real investigations because multiple devices rarely share a perfect clock. If the camera time is off by even a few seconds relative to radar, ADS-B, or another camera, the attempted match-up can fail, and a normal aircraft becomes an “unknown” purely because the timelines don’t line up.

Two basics separate “interesting” from “actionable.” First is time sync: you want a consistent, auditable timeline across every sensor involved. Second is chain-of-custody and security for data and metadata, which is normal investigative hygiene: who captured the file, how it was transferred, what was altered (ideally nothing), and where the original lives. If you can’t show that, you can’t rule out accidental edits, missing context, or simple file handling damage.

This is the same logic behind disciplined data management and reproducible analysis: keep the original data, keep the context, and document handling so someone else can re-check your work and reach the same conclusion.

• Original file: not a screen recording or social-media reupload.
• Exact time: time zone, clock source, and any known offsets.
• Location: lat/long (or nearest landmark) and altitude if available.
• Sensor settings: zoom, focal length, exposure, frame rate, IR mode, filters.
• Platform data: sensor orientation, heading, speed, and maneuvers during capture.
• Calibration status: last calibration date, health flags, and any known limitations.
• Second measurement: another sensor type or another viewpoint with matching timestamps.
• Custody trail: who handled the file, how it was transferred, and what processing occurred.

People aren’t irrational for suspecting a cover-up. Classified sensors exist, and when a clip is released without context, it feels like someone is withholding the “real” data. The report’s stance is less dramatic and more stubborn: most blockage is evidentiary quality, not proof of hidden alien tech. If the best available record is uncalibrated, single-sensor, and missing metadata, the outcome is predictable: “unidentified,” not “identified as ET.”

The actionable way to read new UAP news in 2025 to 2026 is to think like an investigator: before you share a clip or accept a strong conclusion, ask for the time, place, sensor settings, calibration status, and a second measurement. If those basics aren’t there, the most honest label is the boring one: insufficient data.

AI Tools NASA Wants Next

If the bottleneck is data quality and comparability, the next question is what helps at scale. That’s where the report’s AI emphasis fits: not as a reveal mechanism, but as a way to handle volume once the fundamentals are in place.

AI won’t reveal aliens, but it can drastically improve how fast we sort signal from noise. NASA’s positioning is straightforward: AI is a force multiplier for triage and pattern-finding, not a magic “alien-detector.” The catch is just as important as the promise. If your data pipeline isn’t standardized, AI doesn’t scale insight, it scales confusion.

NASA’s framing starts with the boring stuff on purpose: common data formats, reproducibility expectations, and consistent collection practices. That’s the only way you can compare one incident to another without arguing about what the sensors were, how they were calibrated, or what the file actually represents.

NASA already has institutional muscle for this kind of data plumbing. The agency’s Earth Science Data Systems (ESDS) Program creates standards to promote interoperability across NASA Earth science data systems, which is exactly the mindset a serious UAP workflow needs: interoperable data you can merge, reprocess, and audit across teams and time (ESDS program).

On the policy side, Earthdata provides guidance for an Open Science Data Management Plan (OSDMP) aligned with NASA open-science expectations and NASA’s scientific data management policy. See Earthdata’s guidance on data management plans and NASA’s Scientific Data Management Policy for related requirements (Earthdata OSDMP guidance; NASA Scientific Data Management Policy (SPD-41)). In practice, that means a credible “AI-enabled UAP” effort should publish how data is stored, described, versioned, and shared so other analysts can reproduce results instead of re-litigating them.

Once the data is standardized, NASA’s “AI tools” idea becomes concrete: analytics that help humans focus. sensor fusion is combining multiple sensors/sources into a more reliable estimate, which lets a UAP workflow reconcile imperfect viewpoints (for example, video plus timestamps plus positional telemetry) into one coherent track you can sanity-check.

anomaly detection is flagging unusual patterns, which is ideal for UAP workflows because it can automatically surface “this deviates from baseline” segments in large volumes of radar-like or telemetry-like time series so a human analyst can validate what’s real versus what’s instrument behavior.

This isn’t speculative capability for NASA. NASA uses AI to analyze satellite datasets and detect anomalies, and it also applies AI in operational contexts like rover planning and scheduling. The point isn’t hype, it’s readiness: the agency already treats AI as an operational tool for classification, quality control, and exception handling.

The research framing lines up with that operational posture. NASA-facing AI/ML discussions explicitly include application areas like “Fault Detection and Analysis,” along with classification-oriented workflows and anomaly detection concepts (including active learning to improve models as new labeled cases come in). That maps cleanly to UAP work: most cases are mundane, a small subset are ambiguous, and you want the system to get better at separating the two.

Computer vision on video: AI can stabilize and label video frames, detect objects, and estimate apparent motion. That’s useful for triage and for building consistent annotations. The friction is that video is often heavily compressed, cropped, or missing sensor context, so AI can confidently “detect” compression artifacts or tracking glitches. Treat the output as a lead, not a verdict.

anomaly detection on time series: Feed radar-like returns, transponder gaps, IMU-like sensor streams, or tracking residuals into an anomaly detector and it will flag segments that depart from baseline behavior. You get faster review and better prioritization, but you also get false positives any time the baseline is wrong (weather, sensor mode changes, maintenance states). Humans still do the adjudication.

Probabilistic decision support: A model can score and rank hypotheses such as “most consistent with balloon,” “aircraft,” “atmospherics,” or “unknown,” based on features you define and document. Done right, this produces an auditable shortlist with error bars. Done wrong, it turns hidden assumptions into authoritative-sounding numbers.

The common failure mode is simple: garbage in, garbage out. Missing metadata and calibration details don’t just reduce accuracy, they make the result impossible to interpret.

If you want to judge any AI-flavored UAP program, demand the basics that make the outputs trustworthy:

1. Standardize the data and metadata so cases are comparable.
2. Disclose methods, model versions, and preprocessing so results are reproducible.
3. Quantify uncertainty with error bars and clear limits on what the model saw.
4. Validate against known objects and known sensor failure modes before chasing “unknowns.”

Rule of thumb for headlines: if the story can’t tell you what data went in, how it was standardized, and how uncertainty was measured, the “AI found something” claim is just noise with better branding.

Disclosure Pressure In Congress

All of that technical talk exists in the same ecosystem as hearings, draft language, and public suspicion-so it’s not surprising the two collide. The report is written like a methods memo, while the public conversation often treats the topic like a courtroom drama.

Politics can force attention; it can’t manufacture evidence. That’s why the disclosure conversation keeps colliding with NASA’s 2023 report: Congress and the public are arguing about secrecy, incentives, and trust, while the report speaks in the colder language of methods, evidence thresholds, and data quality. When those two languages meet, people talk past each other because “tell us what you know” and “show us what the data can support” are not the same request.

That mismatch also explains the search behavior. People aren’t typing “data roadmap for UAP research.” They’re typing alien disclosure, government UFO cover-up, non-human intelligence, because hearings and draft legislation train everyone to expect a reveal, not a research program. Public attention rises, political stakes rise, and the implied promise becomes: if officials are holding events about it, there must be something definitive to disclose.

The House Oversight Committee UAP hearing took place July 26, 2023 (10:00 a.m.) in 2154 Rayburn. Testimony included David Grusch, described as a whistleblower and former DoD employee, and former military pilots. That setting matters because it’s built for accountability and narrative, not lab-grade verification.

The complication is that a hearing record can contain serious allegations without containing the underlying materials needed to verify them publicly. Media summaries highlighted claims like the government being “absolutely” in possession of UAPs. Even if a claim is delivered under oath, it’s still testimony until it’s corroborated by documents, chain-of-custody details, and independently checkable data releases. Hearings are great at surfacing process questions, who knew what and when, which offices handled reports, which channels existed, and where oversight failed. They are not a substitute for verifiable datasets.

So the practical way to read that July 26 hearing is as a forcing function. It raised the cost of ignoring the topic and put reporting pathways and classification practices under a spotlight. What it did not do is “prove aliens” or “prove a cover-up” on its own, because proof lives downstream in evidence you can audit.

The Schumer/Rounds UAP Disclosure Act amendment is explicit about its purpose: to provide for the expeditious disclosure of UAP records. It also includes a clear framing line: “All Federal Government records concerning unidentified anomalous phenomena should carry a presumption of immediate disclosure.”

The catch is that legislative language is a policy proposal, not a finding of fact. A presumption of disclosure tells you what some lawmakers want the default to be: disclose unless there’s a specific, defensible reason not to. That’s meaningful for expectations, because it shifts the debate from “should the government share anything” to “what, exactly, justifies withholding this specific record.” But it still doesn’t authenticate any particular allegation. It’s a process lever aimed at records collection, review, and release, not a guarantee of what those records will contain.

The Whistleblower Protection Act of 1989 (as amended) prohibits retaliation against many federal employees who disclose wrongdoing. Dodd-Frank expanded certain whistleblower protections. That framework helps explain why more people come forward during high-salience moments, but it’s not a promise that every claim is true, or that every individual is protected in every situation. These laws set guardrails and processes; they don’t replace corroboration.

1. Separate the buckets: treat hearing testimony as allegations, treat legislative text as proposed rules, treat released records and datasets as evidence.
2. Track primary documents: save the hearing video/transcript, the exact bill or amendment text, and any official document releases with dates and version history.
3. Demand corroboration: look for multiple independent sources (documents, data, or named custodians), not a single compelling narrative.
4. Follow timelines, not clips: write down who claimed what, when they claimed it, and what supporting material was promised or produced.
5. Keep standards consistent: the surge in “UFO sightings 2025” and “UFO sightings 2026” interest is exactly why you should reward releases you can verify, not stories that only escalate.

Signals That Would Change The Story

So what would count as real movement, beyond louder claims or bigger headlines? The report’s answer is basically: show your work, improve the pipeline, and make progress measurable.

If the UAP story is going to change, it’ll change because the process gets measurably better, not because one viral video racks up views or one dramatic claim dominates a news cycle. The “receipts” look boring on purpose: metrics, repeatable datasets, consistent release rules, and public interfaces that let outsiders verify what insiders say.

Start with public throughput. The Department of Defense reported AARO had reviewed over 1,600 cases as of June 1, 2024 (AARO public updates). That number matters less as a headline and more as a baseline: if the system is improving, you should see stable reporting about case-closure rates, clear categories of explanations, and a consistent definition of what “unresolved” means across updates.

AARO has also stated that an unresolved report still contributes to historical and locational trend analyses. That’s a real process signal: “unresolved” shouldn’t be a junk drawer. Watch for trend outputs you can sanity-check, like location clustering, altitude bands, sensor type breakdowns, and whether “unresolved” shrinks because data quality improves, not because labels change.

Release drama usually crashes into paperwork. Declassification is “the authorized change in the status of information from classified information to unclassified information,” and that authorization step is why “just release everything” moves slower than people expect.

NASA’s lane is explicitly procedural: under 14 CFR part 1203, NASA’s Office of Protective Services determines whether requested information may be declassified under the declassification provisions of that part. Read that as a pathway with constraints, not a promise of what you’ll get or when.

NASA’s follow-through is supposed to ride on existing machinery, not vibes. NASA will execute a three-year data strategy with annual implementation plans that include initiatives and metrics to measure progress. Those metrics are exactly the kind of public artifact you can track for real movement.

On the data side, NASA’s Public Access Plan requires proposals or project plans for scientific research funding to include a Data Management Plan (DMP), and NASA’s Scientific Data Management Policy defines roles and responsibilities for research data produced by investigations. Translation: if UAP-related work becomes “normal science,” you should see standardized datasets released in a form other teams can replicate, plus clearer cross-agency data-sharing and cleaner public portals.

• Published case-closure rates, explanation categories, and a stable, explicit “unresolved” handling rule
• Standardized datasets (with metadata) that independent analysts can rerun and reproduce
• Cross-agency data-sharing mechanisms described in plain language, plus a public interface that improves over time
• Higher-level sensor and reporting upgrades: calibration notes, timestamp standards, confidence scoring
• Annual plan metrics you can compare year over year, not one-off press hits

Use this checklist on every big UAP headline in 2025 to 2026: if it doesn’t ship measurable outputs, it’s noise, not disclosure.

A Scientific Roadmap Not A Reveal

Put it all together and the tone of NASA’s 2023 report becomes hard to miss: it’s trying to make the next round of UAP analysis less ambiguous, not more dramatic.

NASA’s UAP report reads like a scientific roadmap: raise data quality, be transparent about what’s knowable, and use modern analytics to sort signal from noise, not a dramatic reveal about non-human intelligence.

The study team reported no evidence that UAP are extraterrestrial in origin, even while acknowledging some incidents remain unexplained. NASA officials were blunt about why: existing data and eyewitness reports are insufficient to reach conclusive determinations about the nature or origin of every case. The lever is tighter collection standards paired with modern analytics, including the AI-enabled approaches the report calls for. Political disclosure pressure is real, but it only moves the needle when it produces verifiable records, not louder claims. NASA also made a concrete institutional move by creating a Director of UAP Research role and appointing Mark McInerney (NASA announcement).

That’s also the clean way to separate “unidentified” from “unexplainable”: if the supporting measurements, metadata, and custody trail aren’t there, the honest conclusion is usually “insufficient data,” not a hidden answer. Stay grounded by tracking evidence-based milestones: improved datasets, published standards, and documented releases you can actually verify.

Frequently Asked Questions