CamoText πŸ›‘ (@CamoText) header
CamoText πŸ›‘ (@CamoText) profile picture

⨳ CamoText: offline text anonymizer ⨳ CamoVoice: offline voice typing and transcription ⨳ CamoConvert: offline file conversion

on device
Joined February 2025
Photo by CamoText πŸ›‘ (@CamoText)Photo by CamoText πŸ›‘ (@CamoText)Photo by CamoText πŸ›‘ (@CamoText)Photo by CamoText πŸ›‘ (@CamoText)Photo by CamoText πŸ›‘ (@CamoText)Photo by CamoText πŸ›‘ (@CamoText)Photo by CamoText πŸ›‘ (@CamoText)Photo by CamoText πŸ›‘ (@CamoText)Photo by CamoText πŸ›‘ (@CamoText)Photo by CamoText πŸ›‘ (@CamoText)
CamoText πŸ›‘ retweeted
Friendly reminder that it’s extemely reckless to be putting sensitive client information into AI. Most AI companies will have humans that read the conversations for quality control of the AI and it could be outsourced (ie the Kenyan contractors that were seeing ALL the footage grabbed by Meta glasses). Any human that sees that info could blackmail you out your clients.
3
5
54
3,154
CamoText πŸ›‘ retweeted
Full article, which includes all prompts used for the report, trial transcripts, the report itself. The firm hired charged more than $80,000 for the report. Also shows that AI transcripts can be discoverable and can be very revealing: 404media.co/show-how-3m-is-0…
19
141
20
1,280
144,239
CamoText πŸ›‘ retweeted
Legal AI dominates headlines. But how are lawyers actually using AI in 2026? To answer that question, I spoke with 10 people at the cutting edge of legal AI. Lawyers like @zackbshapiro, @mollyisonchain, Sujit Raman, @MikeJShowalter, @erichdylus, Aaron Kelly, as well as @CooleyLLP's Chief Innovation Officer, @SpellbookLegal's CEO @scottastevenson and others. By the end of this @LawofCodeFM episode, you'll understand how these lawyers are AI maxxing in 2026. Timestamps: 0:00 Intro 1:58 Why it's the best time to be a lawyer 3:03 Hallucinations and cognitive surrender @zackbshapiro 5:33 Quality over efficiency 8:13 Why AI won't just augment work, it upends it @mollyisonchain 15:20 60% of contracts filed to EDGAR have mistakes @scottastevenson 20:04 How LLMs actually work 25:59 Zero data retention explained 29:16 The privacy risk beyond training @erichdylus 39:16 How to prompt @zackbshapiro 47:03 @MikeJShowalter's AI-native litigation stack 55:27 @SpellbookLegal's Compare to Market Feature 1:03:10 Building a regulatory agent 1:09:50 The judgment crisis for junior lawyers 1:12:15 Cooley's AI training method Thank you to the presenting sponsor of this episode, @altitude. Nothing in this podcast is legal or investment advice.
9
14
4
50
34,912
CamoText πŸ›‘ retweeted
The problem in legal AI was never that lawyers are using it. It's that too many firms are using it with no rule governing how. Above the Law framed it well this month: associates are quietly running client work through AI because they assume everyone else is, lawyers are pasting privileged material into public tools without knowing where it lands, and the written policy is either nonexistent or vague enough to mean "be careful." Here is the part practitioners should sit with. Courts and disciplinary authorities are about to stop treating these as growing pains and start treating them as failures of professional management. A growing pain gets sympathy. A management failure gets a referral. The firms without a real AI policy right now are the ones that will find out what that reclassification costs.
7
4
1
26
2,792
CamoText πŸ›‘ retweeted
πŸ”In legal tech, we often discuss implementing AI into our workflows. But how does that impact privacy? That's the issue @erichdylus solved with @CamoText!
2
2
1
5
736
CamoText πŸ›‘ retweeted
Any lawyers using something like this? How does it affect productivity?
Setting up client PII anonymization in your Claude Cowork process is easier than you think. Claude works just as well, if not better, with explicitly labeled subjects. It’s possible to preserve both output quality and confidentiality! camotext.ai/blogposts/use-cl…
11
2
21
13,171
CamoText πŸ›‘ retweeted
FYI if you have a zdr with anthropic. It doesn't apply to mythos/fable support.claude.com/en/articl…
1
2
146
CamoText πŸ›‘ retweeted
3
14
1
48
3,388
CamoText πŸ›‘ retweeted
Basically every AI prompt is discoverable in litigation, people. Your litigation adversaries implore you to keep vibe lawyering with Claude and ChatGPT Screenshots of the pertinent reasoning
Your AI conversations aren't privileged. Yesterday, Judge Jed Rakoff ruled that 31 documents a defendant generated using an AI tool and later shared with his defense attorneys are not protected by attorney-client privilege or work product doctrine. The logic is simple: an AI tool is not an attorney. It has no law license, owes no duty of loyalty, and its terms of service explicitly disclaim any attorney-client relationship. Sharing case details with an AI platform is legally no different from talking through your legal situation with a friend (which is not privileged). You can't fix it after the fact, either. Sending unprivileged documents to your lawyer doesn't retroactively make them privileged. That's been settled law for years. It just hadn't been tested with AI until now. And here's what really hurt the defendant: the AI provider's privacy policy (Claude), in effect when he used the tool, expressly permits disclosure of user prompts and outputs to governmental authorities. There was no reasonable expectation of confidentiality. The core problem is the gap between how people experience AI and what's actually happening. The conversational interface feels private. It feels like talking to an advisor. But unless you negotiate for an enterprise agreement that says otherwise, you're inputting information into a third-party commercial platform that retains your data and reserves broad rights to disclose it. Judge Rakoff also flagged an interesting wrinkle: the defendant reportedly fed information from his attorneys into the AI tool. If prosecutors try to use these documents at trial, defense counsel could become a fact witness, potentially forcing a mistrial. Winning on privilege doesn't make the evidentiary picture simple. For anyone advising clients or managing legal risk, this is a wake-up call. AI tools are not a safe space for clients to process their counsel's advice and to regurgitate their legal strategy. Every prompt is a potential disclosure. Every output is a potentially discoverable document. So what do we do about it? First, attorneys need to be proactive. Advise clients explicitly that anything they put into an AI tool may be discoverable and is almost certainly not privileged. Put it in your engagement letters. Make it part of onboarding. Don't assume clients understand this, because most don't. Second, if clients want to use AI to help process legal issues (and they clearly will, increasingly), then let's give them a way to do it inside the privilege. Collaborative AI workspaces shared between attorney and client, where the AI interaction happens under counsel's direction and within the attorney-client relationship, can change the analysis entirely. I'm excited to be planning this kind of approach, and I think it's where the industry needs to head. storage.courtlistener.com/re…
3
4
1
21
5,730
CamoText πŸ›‘ retweeted
BREAKING: @OpenAI must tuner over 20 million+ chat logs to plaintiffs, Judge Ona Wang has ruled in a 9-pg Order just issued:
91
82
165
471
1,129,439
Holy moly, read this entire thread before using AI browsers. Scary stuff. Fantastic research from the @brave team. TLDR The AI tools you use matters a HUGE deal when it comes to your privacy and security, choose carefully.
Replying to @brave
Perplexity Comet’s AI assistant can take screenshots of websites and analyze them for users. However, Comet will also follow instructions hidden on a webpage that it screenshots.
30
374
8
1,714
111,704
CamoText πŸ›‘ retweeted
Masked, not erased: how broken redaction fuels AI data leaks buff.ly/OtJib23
4
3
80
CamoText πŸ›‘ retweeted
Privacy is normal, and privacy should be preferred. As LLMs improve and compute continues to become cheaper, data collection will increase until our lives and personalities can be remotely reconstructed and commodified. Compliance creep has led non-financial businesses to implement KYC obligations that aren’t legally required nor in their customers’ interest. The sweeping databases of personal and financial data have found far more use for identity thieves than for legitimate law enforcement purposes. Privacy policies are updated server-side, and lengthy terms are rarely read. In addition to astronomical compliance expenses and data breach harms, there are extensive chilling effects on innovation and economic growth. However, better LLMs and cheaper compute also means that on-device application capabilities will expand, and the barrier to entry for creators will reduce. We need to build and use privacy tools where possible that process data locally and client-side (local LLMs and P2P software), avoid intermediary rent-seeking and censorship (smart contract payment flows) and reveal only the satisfaction of a requirement rather than the applicable characteristic itself (zk proofs), while we still have some aspects of ourselves to ourselves. (link below)
2
2
13
839
CamoText πŸ›‘ retweeted
First real AI data breach. Loads of people doxxed doing outrageous shit with chatGPT. Did you know as soon as you hit the share button it creates a link indexed by Google and everyone in the world can read that chat? No? Neither did they clearly.
ChatGPT quietly scrubbed today nearly 50,000 shared conversations from Google's index after our investigation. They thought they'd solved the problem. They were wrong. (1/5)
17
14
8
109
16,369
CamoText πŸ›‘ retweeted
End-to-end encryption secures data in transit, but once it lands, privacy depends on the device's integrity. This underscores the importance of maintaining strong privacy practices at the device level.
2
5
22
597
Regulations change with politics. Encryption doesn’t. If your privacy depends on a law, it can be repealed. If it depends on math, it holds, no matter who’s in power. Choose tools that are private by design, not just by promise.
24
113
12
520
24,161
CamoText πŸ›‘ retweeted
Saying privacy is already lost overlooks the progress being made in developing practical, privacy-preserving tools. This view risks discouraging participation in efforts that are still very much worth pursuing - now more than ever. Let's band together and usher in a new era of freedom tech.
4
10
1
32
1,504
After some tool calls, Claude creates a PR, leaking everything. This includes: full names, travel plans, salary, list of private repos, etc.
2
6
1
55
21,475
CamoText πŸ›‘ retweeted
No better advertisement for privacy tooling than β€œif our LLM thinks you’re doing something bad it will call the authorities on you”
3
5
14
1,181
CamoText πŸ›‘ retweeted
Privacy begins with awareness, not perfection. Every online activity produces data that is often shared without explicit consent. Recognizing that is the first step. The next step is to take deliberate measures to reduce the amount of personal data you share.
2
8
184