What most AI adoption stories miss is the human side effect
We see 2 things happening right now
Authenticity Penalty: when people get something that looks AI generated they just value it less. Polished mails, long pull requests, flawless support replies - the moment it smells like AI trust goes down
Synthetic Fatigue: people get tired of dealing with synthetic voices, support bots, canned messages. They are polite, balanced, verbose - but people are simply fed up with it
This is not about if AI output is <good enough>. It is about the perception of effort and authenticity. Humans notice if someone really invested time. Even flawed writing can feel more real than machine perfection.
It’s a bit like the Japanese idea of wabi sabi. The imperfection is what makes something real and worth something.
So I think we will see a counter movement. A return to human made interaction. People will wait longer or pay more just to avoid synthetic exchanges
AI can scale content but it cannot scale authenticity - And that gap is starting to matter.
Witkoff: "There's only one thing I wish for: that that Nobel committee finally gets its act together and realizes that you are the single finest candidate since that Nobel award was ever talked about."
Before moving from my role at Google to Snowflake I sat down and did a braindump of all the guidelines that I follow (or followed at one point and wanted to reintroduce).
For those interested, here are the ~34 guidelines that made the cut
We analyzed the top 500 most successful THOR rules – “successful” meaning: they detected samples that were either ignored or missed by nearly all AV engines on VirusTotal.
Some rules detect clear malware. Others reveal dual-use tools, renamed hacktools, misused admin binaries, or forensic leftovers.
Most of these samples showed 0 AV detections, the rest only minimal hits.
Not all threats are payloads. Not all detections are flashy.
But these rules consistently light up the blind spots in AV and EDR coverage – where attackers hide comfortably.
THOR doesn’t replace existing tools. It shows you what they forgot to tell you.
https://t.co/PlcVf2n8qE
I updated the slide on common entry vectors.
Revised some wording, added realistic examples (Exchange, Ivanti, …) and simplified countermeasures.
Still not trying to cover everything – just the obvious stuff.
Box size still roughly reflects prevalence.
Feedback welcome.
@SWagenknecht@SWagenknecht, Sie sind so erbärmlich. Nehmen Sie einfach Ihre 4,98% und verschwinden Sie auf Nimmerwiedersehen im Saarland oder ziehen Sie nach #Russland
@RMVdialog könntet Ihr an einem Streiktag der U-Bahnen, an dem jede zweite S2 zusätzlich ausfällt, die stattfindenden Fahrten dann nicht ggf. wie üblich mit 3 Zugteilen ausstatten? Wer entscheidet „Ach, heute reichen 2!“ S2 Taunusanlage 8:18
3/ von den im Bundestag vertretenen 80% die Hälfte für Gesetzesvorhaben mobilisieren. Also 40%-Punkte von insgesamt 60 auf sie entfallenden Prozenten. Das ist eine 2/3-Mehrheit. Bedeutet das bei den unterschiedlichen Positionen dann Stillstand???
Mathematik macht vieles deutlich.
Viel ist davon die Rede, dass Mehrheiten im Bundestag ohne die AfD zu organisieren seien. Wieviele Bürger haben dabei erkannt, dass dies bedeutete, Entscheidungen quasi nur noch mit verfassungsändernden Mehrheiten erreichen zu können?
2/ Ziehen FDP, BSW, Linke sowie Kleinparteien nämlich mit insgesamt ca. 20% gar nicht in den Bundestag ein und erreicht die AfD die prognostizierten 20%, so müssten die demokratischen Parteien...
#100DaysOfYara
Everyone can write YARA rules
Stand out from the crowd!
Aim for:
- Robustness: Keep FPs low
- Scope: Target entire families or techniques, not just specifics
- Efficiency: Avoid regex, loops, math, & hashing
- Clarity: Include detailed descriptions & metadata