Provenance record

Cyber criminals use AI to supercharge scams, watchdog warns

Financial Times: Artificial Intelligence (tier 1, news) 2026-09-14T04:00:12.000Z Original ↗

Source note: Headlines and standfirsts only; full text is paywalled.

Discourse valence

-64

Adverse

confidence 73% · 5 items · range -100 to -40

-100

-50

+50

+100

-64

Adverse readingFavourable reading

Consensus of 5 models from different labs. Spread 60 points, agreement low.

Existential and catastrophic risk Loss of control and alignment Concentration of power

Excerpt as ingested

Criminals are switching from using vast fraud compounds with hundreds of workers to one or two people working on sophisticated models

Every model that read this

Model	Provider	Stage	Score	Conf.	Latency	Prompt	When
Llama 3.3 70B	Meta	analysis	-75	80%	4308ms	v1.0.0 / m1.0.1	2026-09-14 05:41
GPT-4.1 mini	OpenAI	consensus	-40	80%	2886ms	v1.0.0 / m1.0.1	2026-09-14 05:42
Claude Sonnet 5	Anthropic	consensus	-45	45%	5151ms	v1.0.0 / m1.0.1	2026-09-14 05:42
Gemini 2.5 Flash	Google	consensus	-100	80%	2615ms	v1.0.0 / m1.0.1	2026-09-14 05:42
Mistral Small 3.1 24B	Mistral AI	consensus	-60	80%	5736ms	v1.0.0 / m1.0.1	2026-09-14 05:42

Llama 3.3 70B · reading

AI used for scams

evidence: reported horizon: n/a

GPT-4.1 mini · reading

The claim indicates criminals using AI to enhance scams, reducing the need for many workers to just a few with sophisticated models. This implies increased efficiency and potential escalation in fraud risks, negatively impacting society. While the scale and specifics are not detailed, the implication is materially concerning.

evidence: reported horizon: near capability 50 societal 0 existential 0 economic 0

Claude Sonnet 5 · reading

The item reports a watchdog warning, not measured statistics, about AI enabling smaller groups to run scams previously requiring large compounds. This is a real but somewhat speculative claim about criminal efficiency gains, with limited detail given the paywall.

evidence: speculative horizon: n/a

Gemini 2.5 Flash · reading

The article states that criminals are using AI to 'supercharge scams', moving from large compounds to small teams of one or two people operating sophisticated models. This implies a significant increase in the efficiency and scale of criminal activity, enabling more harm with fewer resources. The implication for humanity is materially concerning due to increased fraud.

evidence: reported horizon: n/a

Mistral Small 3.1 24B · reading

The article warns of AI being used to enhance criminal activities. This implies a negative impact on society. The evidence is speculative, as it is a warning from a watchdog.

evidence: speculative horizon: n/a

Evidence extracted

Criminals use AI to supercharge scams
Criminals switch to sophisticated models with fewer workers

The chain

SOURCE     Financial Times: Artificial Intelligence (tier 1)
   ↓
DOCUMENT   50e8c7ff-abf2-4b41-8ae7-7b01695887ee
           https://ft.com/content/d545bae1-d770-46f3-9913-07e85d4ae34c
   ↓
EVIDENCE   2 extracted excerpts
   ↓
MODEL RUN  5 runs, methodology 1.0.1
   ↓
SCORE      -64  (Adverse)
   ↓
CONFIDENCE 73%