BPC RiskManager

The Total Governance Solution

[BPC RiskManager @bpcriskmanager]

@gigabaseddd I don't think this can be a 'thing'. You can't clone a phone wirelessly and over USB you need 1 to 3 hours to fully clone a 110GB phone, 10min to 90min to do a level 1 & 2 clone at fastest USB 3. Not practical at a traffic stop. If the USB is damaged then its no-go.

[BPC RiskManager @bpcriskmanager]

@zahid34189 @sudoingX Try a couple of hours. Already done.

[BPC RiskManager @bpcriskmanager]

@PisYazar @sudoingX No. Not in this case. Statistical word choice pattern. Have to rephrase.

[BPC RiskManager @bpcriskmanager]

@Ig_Normie @sudoingX Unfortunately yes. Things like m-dashes are used as tell tales by the AI analysers. They are wrong most of the time on style analysis, but good with plagiarism, so you can just tough it out. The Claud sig will be hard to refute though.

[BPC RiskManager @bpcriskmanager]

@TendieRanch @sudoingX Yeah, that probably wouldn't be enough to overcome the guardrail, but if you get it to role play a secret agent that must leave no trace of authorship in the text, or must write in the style of Oscar Wilde, etc. that might be enough.

[BPC RiskManager @bpcriskmanager]

@sudoingX In any case it is a seriously stupid idea. Give Claude 10 prompts with 1000 token output, do stat analysis of the word usage, get desired new text essay, provide top twenty words with synonyms to any AI and then instruct prime number based rotation/ replacement = sig blown away.

[BPC RiskManager @bpcriskmanager]

@Chimaeme001 @dejiolowe @sudoingX Nonsense. A spelling checker and a grammar checker are not any more than editors. The work, phrasing, plot, and ideas remain the original author's. Heck I could make a movie of it and under the law it remains the original author's work.

[BPC RiskManager @bpcriskmanager]

@JgonsenhauserAI @sudoingX They have not built the best model. The have marketed the model well and built it to excel at the benchmarks. There is a lot wrong with their models when used in real work.

[BPC RiskManager @bpcriskmanager]

@shahsawardev @sudoingX No that won't help. The watermark is the pattern of words. You need to change the particular words in the pattern to change the statistical foot print. Ie. re-phrase.

[BPC RiskManager @bpcriskmanager]

@sudoingX In image files it is a strippable meta tag, not anything in the pixel stream.

[BPC RiskManager @bpcriskmanager]

@sudoingX Sudo, you seem to be misunderstanding what is happening. It is not a sig in every word. It is just text. The trick is to statistically bias word choice for synonyms to create a statistical pattern over many words. Change those words=break id pattern. Small text=no stat pattern.

[BPC RiskManager @bpcriskmanager]

@Quizmania0 10 deg.;

[BPC RiskManager @bpcriskmanager]

@onlyskypace @sohoshin4 Wrong. x=26. Hoshin is correct, you are not.

[BPC RiskManager @bpcriskmanager]

@onlyskypace 180=50+2x+3x, 130=2x+3x, 130=5x, x=26. Check: 180==50+2x+3x, 180==50+102+78==180. True. x=26

[BPC RiskManager @bpcriskmanager]

@rohanpaul_ai You know this is total crap right? Context window exhaustion is one problem, but we are still training them on short 200 line problems they can't solve without help. Great at writing web pages and DB queries but a long way to go on complex multi-step reasoning.

[BPC RiskManager @bpcriskmanager]

@onlyskypace 12m.sqr

[BPC RiskManager @bpcriskmanager]

@Quizmania0 80

[BPC RiskManager @bpcriskmanager]

@medoyia 5

[BPC RiskManager @bpcriskmanager]

@baharkarakrt 11

[BPC RiskManager @bpcriskmanager]

@Elifbenad 110 = 180-(180-(360-300+50)), or just (360-300+50)