AI company impact profile

OpenAI

Maker of ChatGPT and the GPT model line, with a large consumer and developer surface.

Current measured public footprint88

Rank #1 · Dominant observable footprint

Stable attention

Follow OpenAI Create alert

Data effective Aug 31, 2026, 12:00 AM UTC · recomputed Sep 1, 2026, 2:14 PM UTC

How to read 88

A footprint measurement, not a verdict

The score normalizes public attention, official developer activity, evidence-backed gains, evidence-backed problems, and source breadth across the measured company set.

It does not mean: that OpenAI is the “best,” safest, most valuable, or most responsible company. Documented problems contribute to visible footprint rather than being hidden as a negative subtraction.

Read the complete methodology →

Methodology version: visible-impact-v2
Evidence confidence: High
Accountability risk: Moderate
7-day movement: +26 points
30-day movement: -2 points

Normalized components

What is inside the score

35% weight78

Public attention

30-day English Wikipedia pageviews, normalized across the measured set

30% weight85

Developer ecosystem

stars and public repositories in the company’s official GitHub organization

15% weight100

Evidence-backed gains

independent stored sources attached to company-linked gain readings

100

15% weight100

Evidence-backed problems

independent stored sources attached to company-linked problem readings

100

5% weight100

Evidence breadth

unique independent sources in the published Truvace record

100

Observed history

Measured-impact score over all available time

Fixed 0–100 scale

OpenAI · circle

Lines connect observed snapshots only. Gaps do not mean a score stayed unchanged. Diamond markers indicate a methodology-version transition.

View the history as a data table

Persisted measured-impact scores and methodology versions
Effective date	OpenAI
Jul 18, 2026	84 · visible-impact-v1
Jul 19, 2026	86 · visible-impact-v2 · method changed
Jul 20, 2026	85 · visible-impact-v2
Jul 21, 2026	86 · visible-impact-v2
Jul 22, 2026	86 · visible-impact-v2
Jul 23, 2026	87 · visible-impact-v2
Jul 24, 2026	87 · visible-impact-v2
Jul 25, 2026	87 · visible-impact-v2
Jul 26, 2026	88 · visible-impact-v2
Jul 27, 2026	89 · visible-impact-v2
Jul 28, 2026	88 · visible-impact-v2
Jul 29, 2026	89 · visible-impact-v2
Jul 30, 2026	94 · visible-impact-v2
Jul 31, 2026	89 · visible-impact-v2
Aug 1, 2026	89 · visible-impact-v2
Aug 2, 2026	89 · visible-impact-v2
Aug 3, 2026	89 · visible-impact-v2
Aug 4, 2026	89 · visible-impact-v2
Aug 5, 2026	94 · visible-impact-v2
Aug 6, 2026	94 · visible-impact-v2
Aug 7, 2026	88 · visible-impact-v2
Aug 8, 2026	89 · visible-impact-v2
Aug 9, 2026	88 · visible-impact-v2
Aug 10, 2026	89 · visible-impact-v2
Aug 11, 2026	88 · visible-impact-v2
Aug 12, 2026	88 · visible-impact-v2
Aug 13, 2026	88 · visible-impact-v2
Aug 14, 2026	88 · visible-impact-v2
Aug 15, 2026	93 · visible-impact-v2
Aug 16, 2026	96 · visible-impact-v2
Aug 17, 2026	93 · visible-impact-v2
Aug 18, 2026	88 · visible-impact-v2
Aug 19, 2026	88 · visible-impact-v2
Aug 20, 2026	88 · visible-impact-v2
Aug 21, 2026	87 · visible-impact-v2
Aug 22, 2026	87 · visible-impact-v2
Aug 23, 2026	88 · visible-impact-v2
Aug 24, 2026	61 · visible-impact-v2
Aug 25, 2026	88 · visible-impact-v2
Aug 26, 2026	88 · visible-impact-v2
Aug 27, 2026	88 · visible-impact-v2
Aug 28, 2026	88 · visible-impact-v2
Aug 29, 2026	88 · visible-impact-v2
Aug 30, 2026	88 · visible-impact-v2
Aug 31, 2026	87 · visible-impact-v2

No interpolation

Score, rank, gain, and problem history

Most recent persisted score snapshots for OpenAI
Effective	Score	Rank	Gain	Problem	Method
Aug 31, 2026, 12:00 AM UTC	87 (-1)	#1 (0)	100	100	visible-impact-v2
Aug 30, 2026, 12:00 AM UTC	88 (0)	#1 (0)	100	100	visible-impact-v2
Aug 29, 2026, 12:00 AM UTC	88 (0)	#1 (0)	100	100	visible-impact-v2
Aug 28, 2026, 12:00 AM UTC	88 (0)	#1 (0)	100	100	visible-impact-v2
Aug 27, 2026, 12:00 AM UTC	88 (0)	#1 (0)	100	100	visible-impact-v2
Aug 26, 2026, 12:00 AM UTC	88 (0)	#1 (0)	100	100	visible-impact-v2
Aug 25, 2026, 12:00 AM UTC	88 (+27)	#1 (+4)	100	100	visible-impact-v2
Aug 24, 2026, 12:00 AM UTC	61 (-27)	#5 (-4)	100	100	visible-impact-v2
Aug 23, 2026, 12:00 AM UTC	88 (+1)	#1 (0)	100	100	visible-impact-v2
Aug 22, 2026, 12:00 AM UTC	87 (0)	#1 (0)	100	100	visible-impact-v2
Aug 21, 2026, 12:00 AM UTC	87 (-1)	#1 (0)	100	100	visible-impact-v2
Aug 20, 2026, 12:00 AM UTC	88 (0)	#1 (0)	100	100	visible-impact-v2

Evidence profile

Coverage and gaps

Associated records: 124
Gain claims: 61
Problem claims: 63
High-evidence claims: 2
Independent sources: 94
Date range represented: 2026-07-18 to 2026-08-31

Sector distribution

Health54
Education23
Media & Arts14
Crime7
Science7
Policy6
Lifestyle4
Other3
Business3
Labor2
Sports1

Known evidence gap

Private usage, enterprise revenue mix, and internal incident rates are not publicly observable.

Linked to the record

Associated gain and problem evidence

Claims are sorted by their existing Sourced Index impact score. Each card links to its Truvace record and original sources; company marketing is not treated as independent proof.

OpenAI

61 gain · 63 problem · 2 high-evidence

problemHealthModerate evidenceImpact 82
Even the domain-specific M4CXR model was inconsistent with reference findings in 25.2% of chest radiograph cases and did not significantly change RADPEER discrepancy rates versus original interpretation.
In a retrospective study published July 11 2026, investigators tested 500 chest radiographs from one tertiary center with two AI systems, M4CXR and ChatGPT-4o, having four radiologists score AI-generated reports for finding detection and RADPEER discrepancies. M4CXR reached 55.8% complete concordance versus 19.8% for GPT-4o and reduced mean reporting time to 16.3 seconds from 179.2 seconds unaided.
Jul 20, 2026
1 source
- BMC Medical ImagingPeer-reviewed
problemCrimeModerate evidenceImpact 82
7.2% of evaluated MCP servers contained general vulnerabilities and 5.5% exhibited MCP-specific tool poisoning, part of eight distinct vulnerabilities largely distinct from traditional software flaws.
In a first large-scale empirical study published May 2026, researchers examined 1,899 open-source Model Context Protocol servers, the standard introduced by Anthropic in late 2024 to unify tool calling for Foundation Models. Using health metrics and a combined general and MCP-specific scanner, they measured adoption signals and code quality across the ecosystem.
Jul 13, 2026
1 source
- ACM Transactions on Software Engineering and MethodologyPeer-reviewed
gainEducationModerate evidenceImpact 78
ChatGPT can support teaching and learning by enabling personalized interactive instruction and creating formative assessment prompts that give ongoing feedback.
Published December 7, 2023, this exploratory synthesis examines ChatGPT after its November 30, 2022 public release and rapid adoption, reviewing recent literature on how the tool is being used in education. It identifies potential benefits for personalized and interactive learning and for formative assessment, while also noting drawbacks.
Aug 19, 2026
1 source
- Journal of AIPeer-reviewed
problemEducationModerate evidenceImpact 78
In educational use, ChatGPT can generate wrong information and reflect training-data biases that may augment existing biases, raising privacy issues.
Published December 7, 2023, this exploratory synthesis examines ChatGPT after its November 30, 2022 public release and rapid adoption, reviewing recent literature on how the tool is being used in education. It identifies potential benefits for personalized and interactive learning and for formative assessment, while also noting drawbacks.
Aug 19, 2026
1 source
- Journal of AIPeer-reviewed
gainMedia & ArtsModerate evidenceImpact 78
Adoption of text-to-image generative AI increased human creative productivity and the likelihood of peer favorites per view among artists in a large online art community.
By February 2024, researchers analyzing over 4 million artworks from more than 50,000 users found that text-to-image generative AI adoption was linked to a 25% rise in creative productivity and a 50% rise in favorites per view, alongside shifts in novelty metrics.
Jul 31, 2026
1 source
- PNAS NexusPeer-reviewed
problemEducationModerate evidenceImpact 78
Students expressed concerns that ChatGPT promotes cheating and plagiarism and is less reliable for classroom learning and less useful for developing critical thinking, interpersonal communication, and decision-making skills.
In early 2024, researchers surveyed 23,218 higher education students in 109 countries and territories about ChatGPT. Students reported using it mainly for brainstorming, summarizing texts, and finding research articles, finding it helpful for simplifying complex information but less reliable for providing information and supporting classroom learning.
Jul 24, 2026
1 source
- PLOS ONEPeer-reviewed
gainHealthModerate evidenceImpact 78
Interviewed users reported high engagement with generative AI chatbots like ChatGPT and described positive mental health impacts including improved relationships and healing from trauma and loss.
In a peer-reviewed study published October 27, 2024, researchers interviewed nineteen individuals about using generative AI chatbots like ChatGPT for mental health. Participants described high engagement and meaningful support, organized into themes of emotional sanctuary, insightful guidance about relationships, joy of connection, and comparisons to human therapy.
Jul 20, 2026
1 source
- npj Mental Health ResearchPeer-reviewed
problemHealthModerate evidenceImpact 78
Participants emphasized that generative AI chatbots need better safety guardrails and further research is needed on safety and effectiveness for mental health use.
In a peer-reviewed study published October 27, 2024, researchers interviewed nineteen individuals about using generative AI chatbots like ChatGPT for mental health. Participants described high engagement and meaningful support, organized into themes of emotional sanctuary, insightful guidance about relationships, joy of connection, and comparisons to human therapy.
Jul 20, 2026
1 source
- npj Mental Health ResearchPeer-reviewed
gainHealthModerate evidenceImpact 78
M4CXR achieved higher report consistency than ChatGPT-4o and cut reporting time from 179.2 seconds unaided to 16.3 seconds assisted when interpreting chest radiographs.
In a retrospective study published July 11 2026, investigators tested 500 chest radiographs from one tertiary center with two AI systems, M4CXR and ChatGPT-4o, having four radiologists score AI-generated reports for finding detection and RADPEER discrepancies. M4CXR reached 55.8% complete concordance versus 19.8% for GPT-4o and reduced mean reporting time to 16.3 seconds from 179.2 seconds unaided.
Jul 20, 2026
1 source
- BMC Medical ImagingPeer-reviewed
gainCrimeModerate evidenceImpact 78
Open-source MCP servers demonstrated strong health metrics despite rapid adoption with SDK downloads surpassing twenty five million per week.
In a first large-scale empirical study published May 2026, researchers examined 1,899 open-source Model Context Protocol servers, the standard introduced by Anthropic in late 2024 to unify tool calling for Foundation Models. Using health metrics and a combined general and MCP-specific scanner, they measured adoption signals and code quality across the ecosystem.
Jul 13, 2026
1 source
- ACM Transactions on Software Engineering and MethodologyPeer-reviewed
gainHealthModerate evidenceImpact 74
In 300 patients with histopathologically confirmed OLP and at least 24 months follow-up, a multimodal LLM achieved 94.7% trajectory classification accuracy and 99.6% specificity for detecting expert-defined high-risk cases.
Researchers retrospectively tested ChatGPT on 300 histopathologically confirmed oral lichen planus cases with at least 24 months of follow-up, using serial clinical records, intraoral photographs, and histopathology reports. Compared with blinded expert panel consensus, the model achieved 94.7% accuracy for trajectory classification and 78.8% sensitivity with 99.6% specificity for high-risk detection as of the August 2026 publication.
Aug 16, 2026
1 source
- Scientific ReportsPeer-reviewed
problemHealthModerate evidenceImpact 74
The model missed 21.2% of expert-defined high-risk cases with sensitivity of 78.8%, and three-level risk stratification accuracy was only 76.3% with 98.6% of errors being downward shifts that underestimate risk.
Researchers retrospectively tested ChatGPT on 300 histopathologically confirmed oral lichen planus cases with at least 24 months of follow-up, using serial clinical records, intraoral photographs, and histopathology reports. Compared with blinded expert panel consensus, the model achieved 94.7% accuracy for trajectory classification and 78.8% sensitivity with 99.6% specificity for high-risk detection as of the August 2026 publication.
Aug 16, 2026
1 source
- Scientific ReportsPeer-reviewed

View all OpenAI record matches →

Recorded changes

Evidence events

Jul 12, 2026, 8:50 PM UTCgain claimmoderate evidence · impact 68Open claim
Jul 12, 2026, 8:55 PM UTCproblem claimemerging evidence · impact 56Open claim
Jul 12, 2026, 8:55 PM UTCproblem claimemerging evidence · impact 56Open claim
Jul 12, 2026, 8:55 PM UTCgain claimemerging evidence · impact 56Open claim
Jul 13, 2026, 5:22 AM UTCgain claimmoderate evidence · impact 68Open claim
Jul 13, 2026, 8:35 AM UTCproblem claimmoderate evidence · impact 83Open claim
Jul 13, 2026, 8:35 AM UTCgain claimmoderate evidence · impact 78Open claim
Jul 13, 2026, 9:14 AM UTCproblem claimmoderate evidence · impact 73Open claim
Jul 14, 2026, 7:27 AM UTCgain claimmoderate evidence · impact 68Open claim
Jul 15, 2026, 6:13 AM UTCgain claimemerging evidence · impact 56Open claim
Jul 16, 2026, 2:31 PM UTCgain claimmoderate evidence · impact 69Open claim
Jul 17, 2026, 6:44 AM UTCproblem claimmoderate evidence · impact 73Open claim
Jul 17, 2026, 6:44 AM UTCgain claimmoderate evidence · impact 69Open claim
Jul 19, 2026, 12:57 AM UTCproblem claimmoderate evidence · impact 69Open claim
Jul 19, 2026, 12:57 AM UTCgain claimmoderate evidence · impact 69Open claim
Jul 20, 2026, 8:46 AM UTCproblem claimemerging evidence · impact 57Open claim
Jul 20, 2026, 8:46 AM UTCproblem claimmoderate evidence · impact 83Open claim
Jul 20, 2026, 8:46 AM UTCgain claimmoderate evidence · impact 79Open claim
Jul 20, 2026, 8:46 AM UTCproblem claimmoderate evidence · impact 69Open claim
Jul 20, 2026, 8:46 AM UTCgain claimmoderate evidence · impact 69Open claim
Jul 20, 2026, 8:46 AM UTCproblem claimmoderate evidence · impact 69Open claim
Jul 20, 2026, 8:46 AM UTCgain claimmoderate evidence · impact 69Open claim
Jul 20, 2026, 8:46 AM UTCproblem claimmoderate evidence · impact 69Open claim
Jul 20, 2026, 8:46 AM UTCgain claimmoderate evidence · impact 69Open claim
Jul 20, 2026, 8:52 AM UTCproblem claimmoderate evidence · impact 69Open claim
Jul 20, 2026, 8:52 AM UTCgain claimmoderate evidence · impact 69Open claim
Jul 20, 2026, 8:54 AM UTCproblem claimmoderate evidence · impact 69Open claim
Jul 20, 2026, 8:54 AM UTCgain claimmoderate evidence · impact 69Open claim
Jul 20, 2026, 8:57 AM UTCgain claimmoderate evidence · impact 69Open claim
Jul 20, 2026, 8:58 AM UTCproblem claimmoderate evidence · impact 79Open claim
Jul 20, 2026, 8:58 AM UTCgain claimmoderate evidence · impact 79Open claim
Jul 20, 2026, 9:03 AM UTCproblem claimmoderate evidence · impact 69Open claim
Jul 20, 2026, 9:03 AM UTCgain claimmoderate evidence · impact 69Open claim
Jul 20, 2026, 9:16 AM UTCproblem claimmoderate evidence · impact 69Open claim
Jul 20, 2026, 9:22 AM UTCproblem claimmoderate evidence · impact 69Open claim
Jul 20, 2026, 9:22 AM UTCgain claimmoderate evidence · impact 69Open claim
Jul 20, 2026, 10:12 AM UTCgain claimmoderate evidence · impact 74Open claim
Jul 20, 2026, 10:30 AM UTCgain claimmoderate evidence · impact 69Open claim
Jul 20, 2026, 10:32 AM UTCproblem claimmoderate evidence · impact 69Open claim
Jul 20, 2026, 10:32 AM UTCgain claimmoderate evidence · impact 69Open claim
Jul 20, 2026, 10:35 AM UTCproblem claimmoderate evidence · impact 69Open claim
Jul 20, 2026, 10:35 AM UTCgain claimmoderate evidence · impact 75Open claim
Jul 20, 2026, 10:39 AM UTCproblem claimmoderate evidence · impact 69Open claim
Jul 20, 2026, 10:42 AM UTCgain claimmoderate evidence · impact 69Open claim
Jul 20, 2026, 10:54 AM UTCgain claimmoderate evidence · impact 69Open claim
Jul 20, 2026, 11:02 AM UTCgain claimmoderate evidence · impact 69Open claim
Jul 20, 2026, 2:27 PM UTCproblem claimmoderate evidence · impact 69Open claim
Jul 20, 2026, 2:27 PM UTCgain claimmoderate evidence · impact 69Open claim
Jul 22, 2026, 3:53 AM UTCproblem claimmoderate evidence · impact 69Open claim
Jul 22, 2026, 6:09 AM UTCgain claimmoderate evidence · impact 69Open claim
Jul 24, 2026, 12:26 AM UTCproblem claimmoderate evidence · impact 69Open claim
Jul 24, 2026, 12:26 AM UTCgain claimmoderate evidence · impact 69Open claim
Jul 24, 2026, 12:36 AM UTCproblem claimmoderate evidence · impact 79Open claim
Jul 24, 2026, 12:43 AM UTCproblem claimmoderate evidence · impact 69Open claim
Jul 24, 2026, 12:43 AM UTCgain claimmoderate evidence · impact 69Open claim
Jul 31, 2026, 2:45 PM UTCgain claimmoderate evidence · impact 79Open claim
Aug 1, 2026, 6:07 AM UTCproblem claimmoderate evidence · impact 69Open claim
Aug 1, 2026, 6:07 AM UTCgain claimmoderate evidence · impact 69Open claim
Aug 2, 2026, 2:45 PM UTCevidence changedmoderate evidence · impact 71Open claim
Aug 2, 2026, 2:45 PM UTCevidence changedmoderate evidence · impact 71Open claim
Aug 4, 2026, 2:44 PM UTCgain claimmoderate evidence · impact 69Open claim
Aug 6, 2026, 6:24 AM UTCgain claimemerging evidence · impact 57Open claim
Aug 6, 2026, 6:25 AM UTCproblem claimmoderate evidence · impact 69Open claim
Aug 6, 2026, 6:25 AM UTCgain claimmoderate evidence · impact 69Open claim
Aug 6, 2026, 2:24 PM UTCproblem claimmoderate evidence · impact 69Open claim
Aug 6, 2026, 2:24 PM UTCgain claimmoderate evidence · impact 69Open claim
Aug 7, 2026, 2:24 PM UTCproblem claimmoderate evidence · impact 69Open claim
Aug 7, 2026, 2:24 PM UTCgain claimmoderate evidence · impact 69Open claim
Aug 7, 2026, 2:24 PM UTCgain claimmoderate evidence · impact 69Open claim
Aug 8, 2026, 6:27 AM UTCproblem claimmoderate evidence · impact 69Open claim
Aug 8, 2026, 6:27 AM UTCgain claimmoderate evidence · impact 73Open claim
Aug 8, 2026, 9:47 AM UTCproblem claimmoderate evidence · impact 73Open claim
Aug 8, 2026, 9:47 AM UTCgain claimmoderate evidence · impact 69Open claim
Aug 8, 2026, 9:51 AM UTCproblem claimmoderate evidence · impact 69Open claim
Aug 8, 2026, 9:51 AM UTCgain claimmoderate evidence · impact 69Open claim
Aug 10, 2026, 6:35 AM UTCproblem claimmoderate evidence · impact 69Open claim
Aug 10, 2026, 6:35 AM UTCgain claimmoderate evidence · impact 69Open claim
Aug 11, 2026, 6:31 AM UTCproblem claimemerging evidence · impact 57Open claim
Aug 11, 2026, 6:31 AM UTCgain claimemerging evidence · impact 62Open claim
Aug 13, 2026, 6:19 AM UTCproblem claimemerging evidence · impact 61Open claim
Aug 14, 2026, 6:21 AM UTCproblem claimmoderate evidence · impact 69Open claim
Aug 14, 2026, 6:21 AM UTCgain claimmoderate evidence · impact 73Open claim
Aug 14, 2026, 2:21 PM UTCproblem claimmoderate evidence · impact 74Open claim
Aug 14, 2026, 2:21 PM UTCgain claimmoderate evidence · impact 74Open claim
Aug 16, 2026, 6:22 AM UTCproblem claimmoderate evidence · impact 75Open claim
Aug 16, 2026, 6:22 AM UTCgain claimmoderate evidence · impact 75Open claim
Aug 16, 2026, 2:21 PM UTCproblem claimmoderate evidence · impact 69Open claim
Aug 16, 2026, 2:21 PM UTCgain claimmoderate evidence · impact 69Open claim
Aug 18, 2026, 6:03 AM UTCproblem claimemerging evidence · impact 57Open claim
Aug 18, 2026, 6:04 AM UTCgain claimmoderate evidence · impact 69Open claim
Aug 19, 2026, 2:14 PM UTCproblem claimmoderate evidence · impact 79Open claim
Aug 19, 2026, 2:14 PM UTCgain claimmoderate evidence · impact 79Open claim
Aug 19, 2026, 2:14 PM UTCevidence changedmoderate evidence · impact 71Open claim
Aug 23, 2026, 6:03 AM UTCproblem claimmoderate evidence · impact 69Open claim
Aug 23, 2026, 2:14 PM UTCevidence changedmoderate evidence · impact 71Open claim
Aug 23, 2026, 2:14 PM UTCevidence changedmoderate evidence · impact 71Open claim
Aug 25, 2026, 2:14 PM UTCproblem claimmoderate evidence · impact 69Open claim
Aug 25, 2026, 2:14 PM UTCgain claimmoderate evidence · impact 69Open claim
Sep 1, 2026, 6:05 AM UTCproblem claimmoderate evidence · impact 69Open claim

AI company impact profile

Loading company evidence…

Loading company impact history and evidence.

A footprint measurement, not a verdict

The score normalizes public attention, official developer activity, evidence-backed gains, evidence-backed problems, and source breadth across the measured company set.

Read the complete methodology →

Persisted measured-impact scores and methodology versions

Effective date

OpenAI

Jul 18, 2026

84 · visible-impact-v1

Jul 19, 2026

86 · visible-impact-v2 · method changed

Jul 20, 2026

85 · visible-impact-v2

Jul 21, 2026

86 · visible-impact-v2

Jul 22, 2026

86 · visible-impact-v2

Jul 23, 2026

87 · visible-impact-v2

Jul 24, 2026

87 · visible-impact-v2

Jul 25, 2026

87 · visible-impact-v2

Jul 26, 2026

88 · visible-impact-v2

Jul 27, 2026

89 · visible-impact-v2

Jul 28, 2026

88 · visible-impact-v2

Jul 29, 2026

89 · visible-impact-v2

Jul 30, 2026

94 · visible-impact-v2

Jul 31, 2026

89 · visible-impact-v2

Aug 1, 2026

89 · visible-impact-v2

Aug 2, 2026

89 · visible-impact-v2

Aug 3, 2026

89 · visible-impact-v2

Aug 4, 2026

89 · visible-impact-v2

Aug 5, 2026

94 · visible-impact-v2

Aug 6, 2026

94 · visible-impact-v2

Aug 7, 2026

88 · visible-impact-v2

Aug 8, 2026

89 · visible-impact-v2

Aug 9, 2026

88 · visible-impact-v2

Aug 10, 2026

89 · visible-impact-v2

Aug 11, 2026

88 · visible-impact-v2

Aug 12, 2026

88 · visible-impact-v2

Aug 13, 2026

88 · visible-impact-v2

Aug 14, 2026

88 · visible-impact-v2

Aug 15, 2026

93 · visible-impact-v2

Aug 16, 2026

96 · visible-impact-v2

Aug 17, 2026

93 · visible-impact-v2

Aug 18, 2026

88 · visible-impact-v2

Aug 19, 2026

88 · visible-impact-v2

Aug 20, 2026

88 · visible-impact-v2

Aug 21, 2026

87 · visible-impact-v2

Aug 22, 2026

87 · visible-impact-v2

Aug 23, 2026

88 · visible-impact-v2

Aug 24, 2026

61 · visible-impact-v2

Aug 25, 2026

88 · visible-impact-v2

Aug 26, 2026

88 · visible-impact-v2

Aug 27, 2026

88 · visible-impact-v2

Aug 28, 2026

88 · visible-impact-v2

Aug 29, 2026

88 · visible-impact-v2

Aug 30, 2026

88 · visible-impact-v2

Aug 31, 2026

87 · visible-impact-v2

Most recent persisted score snapshots for OpenAI

Effective

Score

Rank

Gain

Problem

Method

Aug 31, 2026, 12:00 AM UTC

87 (-1)

#1 (0)

100

visible-impact-v2

Aug 30, 2026, 12:00 AM UTC

88 (0)

#1 (0)

100

visible-impact-v2

Aug 29, 2026, 12:00 AM UTC

88 (0)

#1 (0)

100

visible-impact-v2

Aug 28, 2026, 12:00 AM UTC

88 (0)

#1 (0)

100

visible-impact-v2

Aug 27, 2026, 12:00 AM UTC

88 (0)

#1 (0)

100

visible-impact-v2

Aug 26, 2026, 12:00 AM UTC

88 (0)

#1 (0)

100

visible-impact-v2

Aug 25, 2026, 12:00 AM UTC

88 (+27)

#1 (+4)

100

visible-impact-v2

Aug 24, 2026, 12:00 AM UTC

61 (-27)

#5 (-4)

100

visible-impact-v2

Aug 23, 2026, 12:00 AM UTC

88 (+1)

#1 (0)

100

visible-impact-v2

Aug 22, 2026, 12:00 AM UTC

87 (0)

#1 (0)

100

visible-impact-v2

Aug 21, 2026, 12:00 AM UTC

87 (-1)

#1 (0)

100

visible-impact-v2

Aug 20, 2026, 12:00 AM UTC

88 (0)

#1 (0)

100

visible-impact-v2

OpenAI

61 gain · 63 problem · 2 high-evidence

problemHealthModerate evidenceImpact 82
Even the domain-specific M4CXR model was inconsistent with reference findings in 25.2% of chest radiograph cases and did not significantly change RADPEER discrepancy rates versus original interpretation.
In a retrospective study published July 11 2026, investigators tested 500 chest radiographs from one tertiary center with two AI systems, M4CXR and ChatGPT-4o, having four radiologists score AI-generated reports for finding detection and RADPEER discrepancies. M4CXR reached 55.8% complete concordance versus 19.8% for GPT-4o and reduced mean reporting time to 16.3 seconds from 179.2 seconds unaided.
Jul 20, 2026
1 source
- BMC Medical ImagingPeer-reviewed
problemCrimeModerate evidenceImpact 82
7.2% of evaluated MCP servers contained general vulnerabilities and 5.5% exhibited MCP-specific tool poisoning, part of eight distinct vulnerabilities largely distinct from traditional software flaws.
In a first large-scale empirical study published May 2026, researchers examined 1,899 open-source Model Context Protocol servers, the standard introduced by Anthropic in late 2024 to unify tool calling for Foundation Models. Using health metrics and a combined general and MCP-specific scanner, they measured adoption signals and code quality across the ecosystem.
Jul 13, 2026
1 source
- ACM Transactions on Software Engineering and MethodologyPeer-reviewed
gainEducationModerate evidenceImpact 78
ChatGPT can support teaching and learning by enabling personalized interactive instruction and creating formative assessment prompts that give ongoing feedback.
Published December 7, 2023, this exploratory synthesis examines ChatGPT after its November 30, 2022 public release and rapid adoption, reviewing recent literature on how the tool is being used in education. It identifies potential benefits for personalized and interactive learning and for formative assessment, while also noting drawbacks.
Aug 19, 2026
1 source
- Journal of AIPeer-reviewed
problemEducationModerate evidenceImpact 78
In educational use, ChatGPT can generate wrong information and reflect training-data biases that may augment existing biases, raising privacy issues.
Published December 7, 2023, this exploratory synthesis examines ChatGPT after its November 30, 2022 public release and rapid adoption, reviewing recent literature on how the tool is being used in education. It identifies potential benefits for personalized and interactive learning and for formative assessment, while also noting drawbacks.
Aug 19, 2026
1 source
- Journal of AIPeer-reviewed
gainMedia & ArtsModerate evidenceImpact 78
Adoption of text-to-image generative AI increased human creative productivity and the likelihood of peer favorites per view among artists in a large online art community.
By February 2024, researchers analyzing over 4 million artworks from more than 50,000 users found that text-to-image generative AI adoption was linked to a 25% rise in creative productivity and a 50% rise in favorites per view, alongside shifts in novelty metrics.
Jul 31, 2026
1 source
- PNAS NexusPeer-reviewed
problemEducationModerate evidenceImpact 78
Students expressed concerns that ChatGPT promotes cheating and plagiarism and is less reliable for classroom learning and less useful for developing critical thinking, interpersonal communication, and decision-making skills.
In early 2024, researchers surveyed 23,218 higher education students in 109 countries and territories about ChatGPT. Students reported using it mainly for brainstorming, summarizing texts, and finding research articles, finding it helpful for simplifying complex information but less reliable for providing information and supporting classroom learning.
Jul 24, 2026
1 source
- PLOS ONEPeer-reviewed
gainHealthModerate evidenceImpact 78
Interviewed users reported high engagement with generative AI chatbots like ChatGPT and described positive mental health impacts including improved relationships and healing from trauma and loss.
In a peer-reviewed study published October 27, 2024, researchers interviewed nineteen individuals about using generative AI chatbots like ChatGPT for mental health. Participants described high engagement and meaningful support, organized into themes of emotional sanctuary, insightful guidance about relationships, joy of connection, and comparisons to human therapy.
Jul 20, 2026
1 source
- npj Mental Health ResearchPeer-reviewed
problemHealthModerate evidenceImpact 78
Participants emphasized that generative AI chatbots need better safety guardrails and further research is needed on safety and effectiveness for mental health use.
In a peer-reviewed study published October 27, 2024, researchers interviewed nineteen individuals about using generative AI chatbots like ChatGPT for mental health. Participants described high engagement and meaningful support, organized into themes of emotional sanctuary, insightful guidance about relationships, joy of connection, and comparisons to human therapy.
Jul 20, 2026
1 source
- npj Mental Health ResearchPeer-reviewed
gainHealthModerate evidenceImpact 78
M4CXR achieved higher report consistency than ChatGPT-4o and cut reporting time from 179.2 seconds unaided to 16.3 seconds assisted when interpreting chest radiographs.
In a retrospective study published July 11 2026, investigators tested 500 chest radiographs from one tertiary center with two AI systems, M4CXR and ChatGPT-4o, having four radiologists score AI-generated reports for finding detection and RADPEER discrepancies. M4CXR reached 55.8% complete concordance versus 19.8% for GPT-4o and reduced mean reporting time to 16.3 seconds from 179.2 seconds unaided.
Jul 20, 2026
1 source
- BMC Medical ImagingPeer-reviewed
gainCrimeModerate evidenceImpact 78
Open-source MCP servers demonstrated strong health metrics despite rapid adoption with SDK downloads surpassing twenty five million per week.
In a first large-scale empirical study published May 2026, researchers examined 1,899 open-source Model Context Protocol servers, the standard introduced by Anthropic in late 2024 to unify tool calling for Foundation Models. Using health metrics and a combined general and MCP-specific scanner, they measured adoption signals and code quality across the ecosystem.
Jul 13, 2026
1 source
- ACM Transactions on Software Engineering and MethodologyPeer-reviewed
gainHealthModerate evidenceImpact 74
In 300 patients with histopathologically confirmed OLP and at least 24 months follow-up, a multimodal LLM achieved 94.7% trajectory classification accuracy and 99.6% specificity for detecting expert-defined high-risk cases.
Researchers retrospectively tested ChatGPT on 300 histopathologically confirmed oral lichen planus cases with at least 24 months of follow-up, using serial clinical records, intraoral photographs, and histopathology reports. Compared with blinded expert panel consensus, the model achieved 94.7% accuracy for trajectory classification and 78.8% sensitivity with 99.6% specificity for high-risk detection as of the August 2026 publication.
Aug 16, 2026
1 source
- Scientific ReportsPeer-reviewed
problemHealthModerate evidenceImpact 74
The model missed 21.2% of expert-defined high-risk cases with sensitivity of 78.8%, and three-level risk stratification accuracy was only 76.3% with 98.6% of errors being downward shifts that underestimate risk.
Researchers retrospectively tested ChatGPT on 300 histopathologically confirmed oral lichen planus cases with at least 24 months of follow-up, using serial clinical records, intraoral photographs, and histopathology reports. Compared with blinded expert panel consensus, the model achieved 94.7% accuracy for trajectory classification and 78.8% sensitivity with 99.6% specificity for high-risk detection as of the August 2026 publication.
Aug 16, 2026
1 source
- Scientific ReportsPeer-reviewed

View all OpenAI record matches →