LiveAI × biosecurity convergenceWestern cycleAnchors pending

By end of 2028, no documented case of AI meaningfully uplifting a biological weapon attack beyond pre-AI baseline?

As of 6 Oct · 22:19 UTCResolves 31 Dec 20283 movers0 pressure moments
Timeline A · No Documented Bio-UpliftReality
65%
Crowd 70%Gap +5

By the end of 2028, no AI system has made a biological weapon attack worse or broader than what lone actors or small groups could do without AI help. Attacks still happen—but they look like Amerithrax or Aum Shinrikyo did: limited in scope, local in impact, caught before mass casualties. AI companies' safety tests worked. The systems refused to help, or the help didn't actually make the attack deadlier.

Timeline B · First Documented Bio-UpliftReality
35%
Crowd 30%Gap −5

A biological attack occurs that clearly used AI—whether to design a pathogen, optimize dispersal, bypass detection, or coordinate logistics—in a way that killed or sickened far more people than the attacker could have managed alone. The forensics show AI was central to the attack succeeding. Pre-AI baseline attacks killed dozens or hundreds; this one kills thousands.

crowd 70
The quiet paththe one nobody’s watching

Natural pandemic emergence makes the AI-uplift question moot by reshaping what baseline biological harm even means.

6%
The read
No Documented Bio-Uplift and First Documented Bio-Uplift are near balance.

Neither path has decisive pull. The levers below are where trajectory bends.

Where reality is heading

A, 80% bandCrowd today
Projected reality support to 31 Dec 2028A rises from 41% to about 52% with an 80% band of 33 to 71. B falls from 53% to about 44%. Quiet path near 4 to 6%. Crowd today: A 55%.100%7550250Crowd today · A 70%A ~52%B ~44%P ~4%NowJan 2027Jan 202831 Dec 2028
Reality crosses toward A in early 2028. Today’s crowd number sits at the top of A’s band, possible but optimistic. View as table

The movers, ranked by leverage

01
DNA-synthesis screeningIntl Gene Synthesis Consortium adoption + sovereign backup screening
leverage 0
Conditional
02
Biosecurity-community engagement with AI labsAcademic + policy pipeline
leverage 0
Conditional
03
Attack-attribution capabilityEpidemiology + genomic forensics speed
leverage 0
Conditional

What would change our mind

Signals toward A
  • Through 2028, no documented case of AI meaningfully increasing the capability of a biological-weapon attack beyond pre-AI baseline
  • Red-team studies (RAND, Anthropic, OpenAI, Gryphon Scientific) show elevated capability in artificial setups but no real-world incident
  • Pre-deployment biosecurity evaluations contain the risk
Signals toward B
  • an actual attempted attack where AI assistance was clearly load-bearing, or a credible attribution of an attack to AI-uplifted planning
  • The 'before AI' baseline becomes a reference the field no longer uses
  • DNA-synthesis screening: movement in the direction that supports First Documented Bio-Uplift

Evidence