Q: What is FBI OSINT AI image analysis? A: FBI OSINT AI image analysis refers to AI tools used in open-source investigations to rapidly determine whether images and videos are authentic and whether the accounts sharing them are genuine. These systems combine visual forensics, authenticity scoring, provenance checks like C2PA, and account behavior signals to provide evidence-based analysis. Q: How does the FBI’s Specialized OSINT Tools IDIQ affect access to these technologies? A: The FBI’s Specialized OSINT Tools multi-award IDIQ creates contract paths that let agencies order FBI OSINT AI image analysis solutions from approved vendors during a five-year ordering period through September 2031. Being on the vehicle allows companies like Cyabra to compete for task orders, but task orders are not guaranteed and depend on funding, competition, and security reviews. Q: What step-by-step workflow should investigators use to check an image or video? A: When using FBI OSINT AI image analysis, investigators should follow a clear workflow: collect and preserve originals and metadata, verify source and context, run authenticity scoring, inspect forensic signals, validate provenance, analyze account behavior, cross-validate with other tools, and document every step. Following these stages enables fast triage and creates repeatable, auditable findings for escalation or case closure. Q: What forensic signals do AI tools look for to detect manipulated media? A: AI tools scan for signs like GAN- or diffusion-generated artifacts, face swaps and splice points, and inconsistencies in noise, lighting, shadows, or motion across frames. They also review metadata and provenance (C2PA and EXIF) and combine these visual signals with account and network context to inform authenticity scoring. Q: How should evidence be preserved to maintain chain of custody when analyzing media? A: Preserve the original file whenever possible, record the URL, timestamp, and user ID, and hash the file before analysis to lock integrity. Keep read-only copies, log every transfer and tool run, and store reports and data in secure, auditable systems to maintain an evidentiary chain of custody. Q: What are the known limitations and safeguards for FBI OSINT AI image analysis? A: FBI OSINT AI image analysis has limits such as false positives and negatives near decision thresholds, artifacts introduced by compression or re-uploads, bias in training data, and privacy concerns when collecting public posts at scale. To mitigate these risks, teams should use multiple methods, set score thresholds and escalation rules in advance, red-team processes with hard test sets, and regularly update models and policies. Q: When should analysts escalate findings or seek specialist review? A: Analysts should escalate to a specialist when scores and forensic signals indicate high risk or when the evidence requires deeper technical or legal review. For mixed or unclear results, add human review and validate with a second tool, while clearly documenting actions and rationale for any escalation. Q: How should findings be documented to ensure reports stand up in court? A: To ensure findings from FBI OSINT AI image analysis hold up in court, reports should state the question (authentic, manipulated, or unknown), list methods and tools used, include key visuals like heat maps and frame grabs, and explain confidence and limits. Reports should also record dates, tool versions, analyst names, and chain-of-custody details such as hashed originals and logged transfers to create an auditable record.