Understanding disinformation early.
BITIG studies public digital content with AI and human review, and makes manipulation risks visible at an early stage.
- Public content
- Filtering
- Early signal
- Human review
Our name
𐰋𐰃𐱅𐰏
bitig — Old Turkic for “inscription, document, book”.
A name for the lasting written word. We ask the same question for the digital age: how do we tell reliable information from fast-spreading manipulation?
The problem
In the digital world, the hardest thing to undo is a lie that has already spread.
Traditional fact-checking starts after content has spread. But disinformation is more than false information: it is manipulation aimed at the specific sensitivities of specific groups.
Spread
How does a claim spread?
A claim rarely reaches the mainstream at once. It usually passes through the same four stages. BITIG looks at the first two, while the spread is still early.
Spark
A claim appears in a small community.
Sensitivity
The claim attaches to a group's fear, its anger or an us-versus-them divide.
Amplification
Accounts share it, reframe it and spread it to new audiences.
Mainstream
The claim reaches a wide audience; this is usually when fact-checks appear.
BITIG focuses here
Traditional fact-checking starts here
Approach
Beyond “true or false”
BITIG asks of every piece of content: whom could it affect, and through which sensitivity?
Public data
Only publicly available content on chosen topics is studied.
Filtering
Duplicates, noise and off-topic content are removed.
AI analysis
Language models screen public content; posts that carry a claim are grouped into narratives and analysed in depth.
Risk signal
Each narrative is scored for its spread, its independent sources, signs of coordination and the sensitivities it plays on. The score does not say whether a claim is true or false.
Human review
Researchers check and correct the machine's description of each narrative. Selected corrections become new examples for the system.
Principles
- Public content only
- No individual profiling
- Minimal data
- Human review before publication
- Sensitivity to context, irony and slang
Project
AI-Supported Disinformation Detection and Analysis System
Run at Ankara University, the project started in September 2026 and runs for 12 months. The system is first tested in a narrow, controlled pilot, then widened step by step.
- Institution
- Ankara University
- Duration
- 12 months · 2026–2027
- Field
- Social sciences
- Department
- Computer Education and Instructional Technology
Work
Analysis system
A research system that shows manipulation risks in public content at an early stage.
Crisis simulations
Controlled scenarios that measure how resilient the information space is to generative AI misuse during a crisis.
Digital awareness
Workshops and meetings that help young people recognise manipulation.
Roadmap
From narrow to wide
The system is first tested in a small, controlled pilot; the scope widens as results are validated.
Now
Pilot
A narrow, controlled pilot at Ankara University, 2026–2027.
Goal
Expansion
More universities, institutions and topics.
Goal
International
Research partners from other countries, and other languages.
Event
Launch event
BITIG project launch
Public presentation of the project
- When
- Where
- Ankara Kent Konseyi (city council)
Hacı Bayram Mah. Atatürk Bulvarı No: 18, Opera – Altındağ / Ankara
Attendance by invitation.
The project was launched on 24 October 2026 at Ankara Kent Konseyi.
Get involved
Be part of this work
Researchers
Joint work on method, data ethics and evaluation.
Institutions
Pilot areas, crisis simulations and in-house workshops.
Students
Digital awareness workshops and volunteering at events.
Questions
Does BITIG decide what is true?
No. BITIG does not label content true or false. Each narrative it shows is a circulating claim, not a verdict; BITIG shows how it spreads and which sensitivities it plays on. Only a cited fact-check gives a truth verdict.
Do you collect my personal data?
Only publicly available content on chosen topics is studied. No individual is profiled, no user is tracked, and data is kept to a minimum.
Is BITIG a company?
No. BITIG is a research project at Ankara University.
Which languages does it cover?
The method is not tied to one language. The pilot starts in a single language; extending it to other languages is the goal.
What does human review mean?
A researcher compares the machine's description of a narrative with its evidence posts, then keeps or corrects it. Review does not decide whether the claim is true.
Contact
Academic and institutional collaboration
Contact us about academic collaboration, institutional meetings or press enquiries.
info@bitig.co