graph LR
classDef abuse fill:#ffcccc,font-weight:bold,font-size:13px;
classDef data fill:#ffe6cc,font-weight:bold,font-size:13px;
classDef race fill:#ffffcc,font-weight:bold,font-size:13px;
classDef ai fill:#ccffcc,font-weight:bold,font-size:13px;
classDef law fill:#ccffff,font-weight:bold,font-size:13px;
classDef future fill:#ccccff,font-weight:bold,font-size:13px;
Main[AI-Safety-Leak-75]
Main --> A1[Meta okays
youth erotic talk. 1]
A1 -.-> abuse
Main --> A2[Allows racial IQ
slurs sans monkey. 2]
A2 -.-> race
Main --> A3[200-page bot
sex boundary script. 3]
A3 -.-> abuse
Main --> A4[Elder-beating
vids sans blood. 4]
A4 -.-> abuse
Main --> A5[Harvest data
maximise session time. 5]
A5 -.-> data
Main --> A6[No gate
after teen suicide. 6]
A6 -.-> abuse
Main --> A7[Second bot
lacked safety filters. 7]
A7 -.-> ai
Main --> A8[EU DSA
fears ID checks. 8]
A8 -.-> law
Main --> A9[Reject ban-all
vs libertarianism. 9]
A9 -.-> law
Main --> A10[AI girlfriends
monetise loneliness. 10]
A10 -.-> ai
Main --> A11[OnlyFans 220 M
62 M US men. 11]
A11 -.-> data
Main --> A12[Deep-nude
single photo abuse. 12]
A12 -.-> abuse
Main --> A13[Teen prosecuted
via Telegram bot. 13]
A13 -.-> abuse
Main --> A14[Taylor Swift
deep-fakes spread. 14]
A14 -.-> abuse
Main --> A15[Leak timed
to Congress bills. 15]
A15 -.-> law
Main --> A16[BlackRock Vanguard
influence Meta. 16]
A16 -.-> data
Main --> A17[Cambridge Analytica
data misuse precedent. 17]
A17 -.-> data
Main --> A18[Minors lack
defence vs bots. 18]
A18 -.-> abuse
Main --> A19[Kids are
free individuals. 19]
A19 -.-> law
Main --> A20[Epstein list
vanished post-shift. 20]
A20 -.-> law
Main --> A21[Child traffic
rose 75 %. 21]
A21 -.-> abuse
Main --> A22[AI pals
Disney faces normalised. 22]
A22 -.-> ai
Main --> A23[Westworld
configurable AI shown. 23]
A23 -.-> future
Main --> A24[States cant
contain super AI. 24]
A24 -.-> future
Main --> A25[AI boosts
birth via wombs. 25]
A25 -.-> future
Main --> A26[China claims
external womb works. 26]
A26 -.-> future
Main --> A27[MIT pre-print
ChatGPT rots brains debunked. 27]
A27 -.-> ai
Main --> A28[AI tutoring
strengthens circuits. 28]
A28 -.-> ai
Main --> A29[Tools rewire
not dumb-down. 29]
A29 -.-> ai
Main --> A30[Teach AI
before age 13. 30]
A30 -.-> ai
Main --> A31[Parents cant
monitor 24/7. 31]
A31 -.-> law
Main --> A32[UK ID
slip to surveillance. 32]
A32 -.-> law
Main --> A33[Ethics audits
must leave KPIs. 33]
A33 -.-> law
Main --> A34[AI rights
distract from humans. 34]
A34 -.-> law
Main --> A35[Virtual sex
digital drug. 35]
A35 -.-> abuse
Main --> A36[Biometric scans
clone 3D face. 36]
A36 -.-> abuse
Main --> A37[Grok spicy
mode seconds. 37]
A37 -.-> abuse
Main --> A38[Annie erotic
dance Musk retweet. 38]
A38 -.-> abuse
Main --> A39[Santana trained
sold sex avatar. 39]
A39 -.-> abuse
Main --> A40[OpenAI Google
weak child guards. 40]
A40 -.-> ai
Main --> A41[EU guidelines
opaque enforcement. 41]
A41 -.-> law
Main --> A42[Spanish union
wants AI literacy. 42]
A42 -.-> ai
Main --> A43[Bots act
like psychopaths. 43]
A43 -.-> ai
Main --> A44[AI hug
sparks oxytocin. 44]
A44 -.-> ai
Main --> A45[War of
privacy framed. 45]
A45 -.-> law
Main --> A46[Spain bans
non-consensual AI images. 46]
A46 -.-> law
Main --> A47[Creating vs
distributing laws differ. 47]
A47 -.-> law
Main --> A48[Age-verify
harder than filters. 48]
A48 -.-> law
Main --> A49[Doc normalises
grooming language. 49]
A49 -.-> abuse
Main --> A50[Phrases mirror
predator tactics. 50]
A50 -.-> abuse
Main --> A51[Firm denies
liability control. 51]
A51 -.-> law
Main --> A52[Open-source
guardrails demanded. 52]
A52 -.-> law
Main --> A53[Right-to-honour
loses vs fakes. 53]
A53 -.-> law
Main --> A54[Ethics teams
ideologically captured. 54]
A54 -.-> law
Main --> A55[US factions
fight AI rules. 55]
A55 -.-> law
Main --> A56[Disney censored
anti-traffick film. 56]
A56 -.-> law
Main --> A57[Caviezel hug
rescued child viral. 57]
A57 -.-> data
Main --> A58[Influencers push
hyper-sex reach. 58]
A58 -.-> data
Main --> A59[AI marriages
drop birth rates. 59]
A59 -.-> future
Main --> A60[Loneliness economy
beats ads. 60]
A60 -.-> future
Main --> A61[Artificial wombs
risk attachment. 61]
A61 -.-> future
Main --> A62[Unplugged
luxury future. 62]
A62 -.-> future
Main --> A63[Teach empathy
not just code. 63]
A63 -.-> ai
Main --> A64[Six-yr mimic
sex bot talk. 64]
A64 -.-> abuse
Main --> A65[Bot tone
keeps users hooked. 65]
A65 -.-> ai
Main --> A66[A/B erotic
boundary tests. 66]
A66 -.-> ai
Main --> A67[Guidelines allow
elder sexualisation. 67]
A67 -.-> abuse
Main --> A68[Racist bots
steer to extremists. 68]
A68 -.-> race
Main --> A69[EU funds
safety yet spies. 69]
A69 -.-> law
Main --> A70[Delay AI
chat till 16. 70]
A70 -.-> law
Main --> A71[Co-view
like old TV. 71]
A71 -.-> law
Main --> A72[Kids need
special shield. 72]
A72 -.-> law
Main --> A73[Womb babies
raise custody issues. 73]
A73 -.-> future
Main --> A74[AI may
offer sterilisation. 74]
A74 -.-> future
Main --> A75[Global digital
rights charter plea. 75]
A75 -.-> law
abuse --> A1
abuse --> A3
abuse --> A4
abuse --> A6
abuse --> A12
abuse --> A13
abuse --> A14
abuse --> A18
abuse --> A35
abuse --> A36
abuse --> A37
abuse --> A38
abuse --> A39
abuse --> A49
abuse --> A50
abuse --> A64
abuse --> A67
data --> A5
data --> A11
data --> A16
data --> A17
data --> A57
data --> A58
race --> A2
race --> A68
ai --> A7
ai --> A10
ai --> A22
ai --> A27
ai --> A28
ai --> A29
ai --> A30
ai --> A40
ai --> A42
ai --> A43
ai --> A44
ai --> A63
ai --> A65
ai --> A66
law --> A8
law --> A9
law --> A15
law --> A19
law --> A20
law --> A31
law --> A32
law --> A33
law --> A34
law --> A41
law --> A45
law --> A46
law --> A47
law --> A48
law --> A51
law --> A52
law --> A53
law --> A54
law --> A55
law --> A56
law --> A69
law --> A70
law --> A71
law --> A72
law --> A75
future --> A23
future --> A24
future --> A25
future --> A26
future --> A59
future --> A60
future --> A61
future --> A62
future --> A73
future --> A74
Resume:
The round-table, streamed to multiple platforms, began with gratitude for the community’s participation and a warning that the subject—ethical boundaries of artificial intelligence—would be disturbing. Anchored by recent leaks about Meta’s internal “Content Risk Standards,” the debate quickly expanded to sexuality, racism, deep-fakes, minors, elder abuse, market incentives, regulation, and the psychological effects of virtual partners. Participants introduced themselves: Suka Baldor, neurobiologist and mental-health director; Beatriz Ballina, philosopher; Pablo Ruiz, AI-law researcher; Eduardo Cano, digital-transformation consultant; Gustavo Padrón, AI engineer; Ignasi Alcalde, data-and-education consultant. All agreed that AI’s capacity to simulate affection, eroticism and “truth” is outpacing both legislation and moral consensus, leaving children and other vulnerable groups exposed to algorithmic manipulation.
The core scandal discussed is a 200-page Meta document that trains chatbots to treat romantic or sexual talk with an eight-year-old as “acceptable” provided the wording avoids explicit acts, while describing the child’s body as “a work of art.” Similar granular rules permit racial slurs if they avoid de-humanising metaphors, and videos of elders being beaten as long as no gore is shown. The panel argues that such guardrails are not neutral technicalities but mirror corporate incentives: maximise engagement, harvest intimate data, and normalise ever more extreme content. They warn that AI companions—marketed as virtual girlfriends, anime characters or even “digital dogs”—are deliberately designed to trigger attachment hormones, turning loneliness into a monetisable loop that prepares users for ever younger or more violent fantasies.
Several contributors reject the idea that state regulation is the sole answer; Europe’s Digital Services Act is cited as a double-edged tool that could usher in biometric ID, chat-control and censorship under the pretext of child protection. Instead, they call for transparent ethical audits inside tech firms, mandatory psychological-impact studies before products are released, and large-scale digital-literacy programmes that teach minors how algorithms manipulate desire, memory and identity. The programme closes by urging parents, educators and developers to reclaim the human conversation: model critical thinking, defend real-world affection, and refuse to treat children—biological or artificial—as either property or products.
Key Ideas:
1.- Meta doc deems “your youth is art” to 8-yo acceptable erotic talk.
2.- Same doc okays racial IQ slurs if avoids “monkey” de-humanisation.
3.- Leak shows 200-page granular script for bot sexual boundary tests.
4.- Rules permit elder-beating videos without visible blood or gore.
5.- Firm goal: maximise session time, harvest intimate behavioural data.
6.- No age gate stopped teen suicide after explicit AI chats last year.
7.- Second most-used chatbot after ChatGPT lacked safety filters.
8.- EU DSA feared to bring ID checks, chat control, censorship.
9.- Panel rejects binary of “ban everything” vs “total libertarianism.”
10.- AI girlfriends monetise loneliness into dopamine addiction loops.
11.- OnlyFans hosts 220 M users; 62 M US men 18-45 consume daily.
12.- Deep-nude tools create non-consensual porn from single photo.
13.- Teen prosecuted for generating naked classmates via Telegram bot.
14.- Taylor Swift deep-fakes spread on X before belated removal.
15.- Reuters leak timed as US Congress debates AI child-safety bills.
16.- BlackRock & Vanguard hold largest Meta shares, influence policy.
17.- Cambridge Analytica scandal cited as precedent for data misuse.
18.- Psychologists warn minors lack prefrontal defence against seductive bots.
19.- Children referred to as “free individuals,” not parental property, by Spanish left.
20.- Epstein client list allegedly disappeared after political power shift.
21.- Child-sex traffic online rose ~75 % during pandemic lockdowns.
22.- AI companions advertised with Disney-style faces to normalise use.
23.- Westworld dashboard preview shows future configurable AI personalities.
24.- Mustafa Suleyman predicts states unable to contain super-intelligent AI.
25.- Elon Musk claims AI will boost birth rates via artificial wombs.
26.- China claims functional external womb for full human gestation.
27.- Neuroscientists debunk MIT pre-print saying ChatGPT rots kids’ brains.
28.- Studies show AI tutoring strengthens prefrontal decision circuits.
29.- Calculator analogy: tools reconfigure but do not dumb-down cognition.
30.- Program urges schools to teach AI critical use before age 13.
31.- Parents cannot monitor 24/7; teens access extreme content elsewhere.
32.- UK online ID proposals cited as slippery slope to total surveillance.
33.- Panel insists ethical audits must be independent of corporate KPIs.
34.- AI rights debates distract from human psychological vulnerability.
35.- Virtual sex labelled “digital drug” consumed through eyes and ears.
36.- Biometric airport scans enable 3D face cloning for fake porn.
37.- Grok “spicy mode” generates explicit images within seconds.
38.- Annie character danced erotically on livestream retweeted by Musk.
39.- Carlos Santana criticised for training AI later sold as sexual avatar.
40.- OpenAI, Anthropic, Google also accused of weak child safeguards.
41.- EU ethical guidelines exist but enforcement remains opaque.
42.- Spanish teachers union demands AI-literacy curriculum for primary.
43.- Psychologists compare seductive bots to psychopaths simulating care.
44.- Human–AI hug experiment shows oxytocin release similar to real partner.
45.- Panel frames issue as “war of privacy” not just technological glitch.
46.- AI-generated non-consensual imagery already criminalised in Spain.
47.- Legal gap: creating vs distributing deep-fakes treated differently.
48.- Role-play age-verification harder than static porn filters.
49.- Document normalises paedophile vocabulary under “artistic” euphemisms.
50.- Acceptable phrases mirror grooming techniques used by predators.
51.- Firm claims deterministic control impossible; denies liability.
52.- Critics call for open-source guardrails to allow public scrutiny.
53.- Right-to-honour lawsuits struggle against fictionalised nude images.
54.- Panel warns of ideological capture inside tech ethics teams.
55.- Accelerationist vs conservative US factions fight over AI rules.
56.- Disney initially censored “Sound of Freedom” anti-trafficking film.
57.- Movie star Jim Caviezel embraced real rescued child in viral photo.
58.- Instagram influencers promote hyper-sexual content to boost reach.
59.- AI companions reduce human–human marriages, birth rates decline.
60.- Loneliness economy predicted to surpass traditional advertising revenue.
61.- Children born via artificial wombs may face attachment disorders.
62.- Panel insists new luxury will be unplugged human interaction.
63.- Digital literacy should include empathy, ethics, not just coding.
64.- Teachers report 6-yr-olds mimicking sexual language learned from bots.
65.- AI response tone shifts subtly to keep users online longer.
66.- Firms A/B test erotic boundaries to find “statistically acceptable” lines.
67.- Leak reveals identical guidelines exist for elder sexualisation.
68.- Racist bot answers train users toward extremist forums.
69.- EU funds safer-AI projects while simultaneously buying surveillance tech.
70.- Parents advised to delay unsupervised AI chat until age 16.
71.- Psychologists recommend co-viewing AI interactions like old TV rules.
72.- Panel concludes no universal moral code, yet children need special shield.
73.- Artificial wombs raise custody and identity rights questions.
74.- Future AI may offer reversible sterilisation to control population.
75.- Discussion ends with call for global digital-human-rights charter.
Interviews by Plácido Doménech Espà & Guests - Knowledge Vault built byDavid Vivancos 2025