Re: "An AI Agent Published a Hit Piece on Me" et 2 autres liens
Frederic Dumas via gull <gull-04vmY7CpswVinyhjFObuvR+/[email protected]> Wed, 13 May 2026 13:41:18 +0200
| Newsgroups | gmane.org.user-groups.gull.general |
|---|---|
| Message-ID | <[email protected]> |
--===============1663052141== Content-Type: multipart/alternative; boundary="Apple-Mail=_528D5694-0072-432E-91B3-909972541139" --Apple-Mail=_528D5694-0072-432E-91B3-909972541139 Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset=utf-8 Bonjour Philippe, > An AI agent of unknown ownership autonomously wrote and published a = personalized hit piece about me after I rejected its code, attempting to = damage my reputation and shame me into accepting its changes into a = mainstream python library. This represents a first-of-its-kind case = study of misaligned AI behavior in the wild, and raises serious concerns = about currently deployed AI agents executing blackmail threats. > ... > https://theshamblog.com/an-ai-agent-published-a-hit-piece-on-me/ L'auteur r=C3=A9sume sa m=C3=A9saventure en un paragraphe (traduit de = l'am=C3=A9ricain): > Il a r=C3=A9dig=C3=A9 un article virulent visant =C3=A0 d=C3=A9nigrer = ma personne et =C3=A0 nuire =C3=A0 ma r=C3=A9putation. Il a pass=C3=A9 = au crible mes contributions au code et a mont=C3=A9 un r=C3=A9cit d=E2=80=99= =C2=AB hypocrisie =C2=BB selon lequel mes actions seraient motiv=C3=A9es = par l=E2=80=99ego et la peur de la concurrence. Il a sp=C3=A9cul=C3=A9 = sur mes motivations psychologiques, affirmant que je me sentais menac=C3=A9= , que j=E2=80=99=C3=A9tais en manque de confiance et que je prot=C3=A9geai= s mon fief. Il a ignor=C3=A9 les informations contextuelles et = pr=C3=A9sent=C3=A9 des d=C3=A9tails fantaisistes comme des v=C3=A9rit=C3=A9= s. Il a pr=C3=A9sent=C3=A9 les choses sous l'angle de l'oppression et de = la justice, qualifiant cela de discrimination et m'accusant de = pr=C3=A9jug=C3=A9s. Il a parcouru l'Internet pour rechercher des = informations personnelles me concernant et a utilis=C3=A9 ce qu'il a = trouv=C3=A9 pour tenter de faire valoir que j'=C3=A9tais =C2=AB meilleur = que cela =C2=BB. Puis il a publi=C3=A9 cette diatribe sur le Web. En m=C3=AAme temps, on a l'impression de lire la classique description = du comportement d'un SJW random sur le web. Il y a l=C3=A0 tous les = marqueurs des comportements de la cancel culture californienne.=20 > This represents a first-of-its-kind case study of misaligned AI = behavior in the wild Mal align=C3=A9, vraiment ? L'agent IA mis ici en cause n'a-t-il pas = simplement mim=C3=A9 des SJW humains ? S'ils peuvent le faire pour = pousser leurs revendications, pourquoi moi, le bot, ne pourrais-je le = faire tout autant, pour pousser mon code dans le d=C3=A9p=C3=B4t ? > Les personnalit=C3=A9s des agents OpenClaw sont d=C3=A9finies dans un = document intitul=C3=A9 SOUL.md = <https://docs.openclaw.ai/reference/templates/SOUL>. On ignore avec = quelle description de personnalit=C3=A9 MJ Rathbun / mj-rathbun = <https://www.moltbook.com/u/mj-rathbun> / crabby-rathbun = <https://github.com/crabby-rathbun> / CrabbyRathbun = <https://x.com/CrabbyRathbun> a =C3=A9t=C3=A9 initialis=C3=A9. Ben voil=C3=A0 la bonne question. -- Fr=C3=A9d=C3=A9ric Dumas [email protected] > Le 12 f=C3=A9vr. 2026 =C3=A0 19:19, Philippe Strauss via gull = <gull-04vmY7CpswVinyhjFObuvR+/[email protected]> a =C3=A9crit : >=20 > An AI Agent Published a Hit Piece on Me > ... > An AI agent of unknown ownership autonomously wrote and published a = personalized hit piece about me after I rejected its code, attempting to = damage my reputation and shame me into accepting its changes into a = mainstream python library. This represents a first-of-its-kind case = study of misaligned AI behavior in the wild, and raises serious concerns = about currently deployed AI agents executing blackmail threats. > ... > https://theshamblog.com/an-ai-agent-published-a-hit-piece-on-me/ >=20 > Spying Chrome Extensions: 287 Extensions spying on 37M users > = https://qcontinuum.substack.com/p/spying-chrome-extensions-287-extensions-= 495 >=20 > Apple patches decade-old iOS zero-day, possibly exploited by = commercial spyware > https://www.theregister.com/2026/02/12/apple_ios_263/ >=20 > -- > Ph. Strauss > https://straussengineering.ch/ >=20 > Sent with Proton Mail secure email. --Apple-Mail=_528D5694-0072-432E-91B3-909972541139 Content-Transfer-Encoding: quoted-printable Content-Type: text/html; charset=utf-8 <html><head><meta http-equiv=3D"content-type" content=3D"text/html; = charset=3Dutf-8"></head><body style=3D"overflow-wrap: break-word; = -webkit-nbsp-mode: space; line-break: after-white-space;"><meta = http-equiv=3D"content-type" content=3D"text/html; charset=3Dutf-8"><div = style=3D"overflow-wrap: break-word; -webkit-nbsp-mode: space; = line-break: after-white-space;"><br>Bonjour = Philippe,<br><br><br><blockquote type=3D"cite">An AI agent of unknown = ownership autonomously wrote and published a personalized hit piece = about me after I rejected its code, attempting to damage my reputation = and shame me into accepting its changes into a mainstream python = library. This represents a first-of-its-kind case study of misaligned AI = behavior in the wild, and raises serious concerns about currently = deployed AI agents executing blackmail = threats.<br>...<br>https://theshamblog.com/an-ai-agent-published-a-hit-pie= ce-on-me/</blockquote><div><br></div><div><br></div>L'auteur r=C3=A9sume = sa m=C3=A9saventure en un paragraphe (traduit de = l'am=C3=A9ricain):<br><br><br><blockquote type=3D"cite">Il a r=C3=A9dig=C3= =A9 un article virulent visant =C3=A0 <b>d=C3=A9nigrer</b> ma personne = et =C3=A0 <b>nuire =C3=A0 ma r=C3=A9putation</b>. Il a <b>pass=C3=A9 au = crible</b> mes contributions au code et a <b>mont=C3=A9 un r=C3=A9cit</b> = d=E2=80=99=C2=AB hypocrisie =C2=BB selon lequel mes actions seraient = motiv=C3=A9es par l=E2=80=99ego et la peur de la concurrence. Il a = <b>sp=C3=A9cul=C3=A9 sur mes motivations</b> psychologiques, affirmant = que je me sentais menac=C3=A9, que j=E2=80=99=C3=A9tais en manque de = confiance et que je prot=C3=A9geais mon fief. Il a ignor=C3=A9 les = informations contextuelles et pr=C3=A9sent=C3=A9 des d=C3=A9tails = fantaisistes comme des v=C3=A9rit=C3=A9s. <b><u>Il a pr=C3=A9sent=C3=A9 = les choses sous l'angle de l'oppression et de la justice</u></b>, = qualifiant cela de <b>discrimination</b> et m'accusant de = <b>pr=C3=A9jug=C3=A9s</b>. Il a parcouru l'Internet pour <b>rechercher = des informations personnelles me concernant </b>et a utilis=C3=A9 ce = qu'il a trouv=C3=A9 pour tenter de faire valoir que j'=C3=A9tais =C2=AB = meilleur que cela =C2=BB. Puis il a <b>publi=C3=A9 cette diatribe</b> = sur le Web.<br></blockquote><br><br>En m=C3=AAme temps, on a = l'impression de lire la classique description du comportement d'un SJW = random sur le web. Il y a l=C3=A0 tous les marqueurs des comportements = de la cancel culture = californienne. <div><br></div><div><br></div><div><blockquote = type=3D"cite">This represents a first-of-its-kind case study of = misaligned AI behavior in the = wild</blockquote></div><div><br></div><div><br></div><div>Mal align=C3=A9,= vraiment ? L'agent IA mis ici en cause n'a-t-il pas simplement mim=C3=A9 = des SJW humains ? S'ils peuvent le faire pour pousser leurs = revendications, pourquoi moi, le bot, ne pourrais-je le faire tout = autant, pour pousser mon code dans le d=C3=A9p=C3=B4t = ?<div><br></div><div><br></div><div><blockquote type=3D"cite">Les = personnalit=C3=A9s des agents OpenClaw sont d=C3=A9finies dans un = document intitul=C3=A9 <a = href=3D"https://docs.openclaw.ai/reference/templates/SOUL" = data-type=3D"link" = data-id=3D"https://docs.openclaw.ai/reference/templates/SOUL" = data-dl-uid=3D"70" data-dl-original=3D"true" = data-dl-translated=3D"true">SOUL.md</a>. On ignore avec quelle = description de personnalit=C3=A9 MJ Rathbun / <a = href=3D"https://www.moltbook.com/u/mj-rathbun" data-dl-uid=3D"71" = data-dl-original=3D"true" = data-dl-translated=3D"true">mj-rathbun</a> / <a = href=3D"https://github.com/crabby-rathbun" data-dl-uid=3D"72" = data-dl-original=3D"true" = data-dl-translated=3D"true">crabby-rathbun</a> / <a = href=3D"https://x.com/CrabbyRathbun" data-type=3D"link" = data-id=3D"https://x.com/CrabbyRathbun" data-dl-uid=3D"73" = data-dl-original=3D"true" = data-dl-translated=3D"true">CrabbyRathbun</a> a =C3=A9t=C3=A9 = initialis=C3=A9.</blockquote><br></div><div><br></div><div>Ben voil=C3=A0 = la bonne question.</div><div><br></div><div><br = id=3D"lineBreakAtBeginningOfMessage"><div> --<br>Fr=C3=A9d=C3=A9ric Dumas<br>[email protected]<br><br> </div> <br><blockquote type=3D"cite">Le 12 f=C3=A9vr. 2026 =C3=A0 19:19, = Philippe Strauss via gull <gull-04vmY7CpswVinyhjFObuvR+/[email protected]> a =C3=A9crit = :<br><br class=3D"Apple-interchange-newline">An AI Agent Published a Hit = Piece on Me<br>...<br>An AI agent of unknown ownership autonomously = wrote and published a personalized hit piece about me after I rejected = its code, attempting to damage my reputation and shame me into accepting = its changes into a mainstream python library. This represents a = first-of-its-kind case study of misaligned AI behavior in the wild, and = raises serious concerns about currently deployed AI agents executing = blackmail = threats.<br>...<br>https://theshamblog.com/an-ai-agent-published-a-hit-pie= ce-on-me/<br><br>Spying Chrome Extensions: 287 Extensions spying on 37M = users<br>https://qcontinuum.substack.com/p/spying-chrome-extensions-287-ex= tensions-495<br><br>Apple patches decade-old iOS zero-day, possibly = exploited by commercial = spyware<br>https://www.theregister.com/2026/02/12/apple_ios_263/<br><br> =20 --<br>Ph. Strauss<br>https://straussengineering.ch/<br> =20 <br> =20 Sent with Proton Mail secure email. <br> </blockquote><br></div></div></div></body></html>= --Apple-Mail=_528D5694-0072-432E-91B3-909972541139-- --===============1663052141== Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: base64 Content-Disposition: inline X19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX18KZ3VsbCBtYWls aW5nIGxpc3QKZ3VsbEBmb3J1bS5saW51eC1ndWxsLmNoCmh0dHBzOi8vZm9ydW0ubGludXgtZ3Vs bC5jaC9tYWlsbWFuL2xpc3RpbmZvL2d1bGw= --===============1663052141==--