| موضوع میں صفحات: < [1 2 3] > | Poll: Does using AI in translation constitute breach of confidentiality? دھاگا پوسٹ کرنے والے: ProZ.com Staff
|
|---|
Lieven Malaise wrote:
Gjorgji Apostolovski wrote:
So if someone wants to spend some of my time, and knowledge to do translation services for them, on which they'll earn big money, then ofc I'll charge them, that's why we are here, right? To submit our applications on offers made by clients. This is not bribery, my friend.
I'll skip the sarcasm next time. Of course translation has got nothing to do with bribery, just like ChatGPT. They are services for God's sake. If you want them, you'll have to pay for them. And yes, promising to not use your input for any other reason than the service itself can be another service for which you have to pay. Nothing weird or wrong about that.
Your sarcasm looked to me like you were comparing translations services with data protection on paid or free versions of software.
This is sarcasm?
"And yes, promising to not use your input for any other reason than the service itself can be another service for which you have to pay."
So loyalty after all has value. Now, this is sarcasm.
So, it's like I'm paying a pro to no sleep with others, but she does anyways.
[Edited at 2025-05-08 18:55 GMT]
| | | | Mario Chávez
امریکہ
Local time: 11:27
رکن (2024)
اسپینیسےانگریزی
+ ...
| The apparent advances in generative IA | May 8 |
Generative IA is the newest shiny toy in town, propelled by a highly self-interested party, Sam Altman et al at Open IA.
The imitation that quickly followed by the likes of Amazon, Microsoft, Apple and Google (to name the main players) shows one of the old-age tropes of capitalism: better buy it now and figure out what to do with it later before the other guys. Irrational? You bet.
And then, the quick adoption of half-baked gen IA products and services ac... See more Generative IA is the newest shiny toy in town, propelled by a highly self-interested party, Sam Altman et al at Open IA.
The imitation that quickly followed by the likes of Amazon, Microsoft, Apple and Google (to name the main players) shows one of the old-age tropes of capitalism: better buy it now and figure out what to do with it later before the other guys. Irrational? You bet.
And then, the quick adoption of half-baked gen IA products and services across different markets and verticals show the degree of stupidity and desperation. Never have I seen such a textbook case of FOMO, the tail wagging the dog.
A potential client/employer last year contacted me to hire me full time to translate educational material (text books) because their clients had begun to come to their senses after experimenting with gen AI. Go figure. ▲ Collapse
| | | | | a little bit off topic | May 8 |
You are off topic, but anyway you forgot to mention that also space company is a thing for some of them with FOMO.
| | | | Daryo
برطانیہ
Local time: 16:27
انگریزیسےسربیائی
+ ...
| And you believe that? | May 9 |
Lieven Malaise wrote:
Payed subscriptions constitute lawful agreements in which is explicitely stated that your data won't be disclosed to third parties or, in the case of DeepL, even will be deleted. This means they are as valid or invalid as the well-known NDAs or other agreements. You can't be sure they will be respected, but they are a legal and therefore acceptable basis to conduct business.
Different story for free versions. The professional use of them is totally unacceptable, but I believe it would be extremely naive to think they aren't widely used by "professional" translators.
Payed subscriptions constitute lawful agreements in which is explicitely stated that your data won't be disclosed to third parties or, in the case of DeepL, even will be deleted.
Well, if you really believe that, you surely still write every year to Santa with your wishes ...
The only 'attenuating circumstance' regarding confidentiality when using any AI is that there are so many other 'weak links' in the communication chain that AIs siphoning any 'training data' they can get their paws on is **only one amongst** many 'holes' leaking information.
| | |
|
|
|
Daryo
برطانیہ
Local time: 16:27
انگریزیسےسربیائی
+ ...
Ahmadreza Karami wrote:
As you know most GAI chatbots which are popular right now are using the new data they receive as a training material which also raises question about copyright issues in some cases which is another topic. Nevertheless, use of AI in case of giving full text or at least some sentences, could be a breach of confidentiality. Of course if you have not signed and NDA and the customer does not require it, you have no legal problem but the fact remains that the text and possibly the translation which you approved after going back and forth with the chatbot is in the dataset of the AI, waiting to be trained on.
But if using the AI is just for understanding small parts of a text, mostly less than a sentence, it could be considered a good enough way to prevent your work leaking to the hand of others for use. From this point forward, it is personal preference whether you want it to be used by others or the public or not.
Of course if you have not signed and NDA and the customer does not require it, you have no legal problem
Well, it DOES NOT work that way.
A serious professional translator / interpreter does not need an explicit reminder in every single contract of his/her duty of confidentiality. It's presumed by default - and if you intend to stay in business (or at least keep the clients worth keeping) you better never forget it.
| | | | Lieven Malaise
بیلجیئم
Local time: 17:27
رکن (2020)
ڈچسےفرانسیسی
+ ...
Daryo wrote:
Well, if you really believe that, you surely still write every year to Santa with your wishes ...
It's not about belief. It's business. They do a legally binding promise to refrain from doing certain things. That's enough in a business environment. Or do you think agencies "believe" that translators won't use AI or machine translation if it's explicitely stated in an NDA or other agreement that they have signed?
[Bijgewerkt op 2025-05-09 05:32 GMT]
| | | | | Deepl Pro Paid Version | May 10 |
The only program/website I use is DeepL PRO (and DeepL WRITE PRO) -paid versions, as they legally state that their data is automatically deleted and never shared with anyone.
https://www.deepl.com/en/pro-privacy_info
I would never use a free version for any translation, as they cannot guarantee confidentiality.
Just in case, I change/omit any personal info... See more The only program/website I use is DeepL PRO (and DeepL WRITE PRO) -paid versions, as they legally state that their data is automatically deleted and never shared with anyone.
https://www.deepl.com/en/pro-privacy_info
I would never use a free version for any translation, as they cannot guarantee confidentiality.
Just in case, I change/omit any personal info (names, personal addresses or any other identifiable information).
[Edited at 2025-05-10 02:56 GMT] ▲ Collapse
| | | | | everything is breachable | May 10 |
bkking wrote:
The only program/website I use is DeepL PRO (and DeepL WRITE PRO) -paid versions, as they legally state that their data is automatically deleted and never shared with anyone.
https://www.deepl.com/en/pro-privacy_info
I would never use a free version for any translation, as they cannot guarantee confidentiality.
Just in case, I change/omit any personal info (names, personal addresses or any other identifiable information).
[Edited at 2025-05-10 02:56 GMT]
Facebook also stated the same, but it was breached.
And it's not just the software itself or some website, you know, there's a lot of ways to breach into someone's data.
So, you can't just say: I paid Pro license and now I can sleep calm.
| | |
|
|
|
Daryo
برطانیہ
Local time: 16:27
انگریزیسےسربیائی
+ ...
| Exercices de style, version 2 | May 10 |
Lieven Malaise wrote:
Daryo wrote:
Well, if you really believe that, you surely still write every year to Santa with your wishes ...
It's not about belief. It's business. They do a legally binding promise to refrain from doing certain things. That's enough in a business environment. Or do you think agencies "believe" that translators won't use AI or machine translation if it's explicitely stated in an NDA or other agreement that they have signed?
[Bijgewerkt op 2025-05-09 05:32 GMT]
Here is a free tip: in order to keep improving, any AI needs continually as much 'training data' as possible.
Businesses declaring 'on their Scout's honour they won't do it' has always stopped businesses from doing it, right? That kind of thing never happens, especially when **it's technically nearly impossible** to check what they are really doing.
I could remind you of a longish list of cases that 'coul never happen', but apparently you have to be 'naive' to even read about them.
Short version: imagine putting within reach of an obsessive-compulsive data-eater some juicy prime-cut data, and expect it to not even touch the data? I surely won’t be doing that with any data I want to keep confidential, any recyclable PR sweet talk of binding legal obligations notwithstanding (BTW are there also 'non-binding' legal obligations ???).
Want to trust confidential data to the kind of people who respond to being caught red-handed pirating other people's data by lobbying to make it legal when THEY do it? Count me out.
| | | |
DeepL (I'm talking about the Pro versions), is subject to the GDPR (General Data Protection Regulation) and the German Federal Data Protection Act (BDSG), which are extremely strict regarding Data Protection. They also highlight advanced encryption and compliance with regulations like ISO 27001 (information security management) and SOC 2 Type II (security, availability, and confidentiality).
By using the Pro services, no texts are stored and are immediately deleted, and the texts a... See more DeepL (I'm talking about the Pro versions), is subject to the GDPR (General Data Protection Regulation) and the German Federal Data Protection Act (BDSG), which are extremely strict regarding Data Protection. They also highlight advanced encryption and compliance with regulations like ISO 27001 (information security management) and SOC 2 Type II (security, availability, and confidentiality).
By using the Pro services, no texts are stored and are immediately deleted, and the texts are not used to improve the quality of their services.
Even then, I find it to be good practice to never include personally identifiable information.
I've found the Pro versions to be a useful aid. Have a nice day everybody.
[Edited at 2025-05-12 23:29 GMT]
[Edited at 2025-05-12 23:30 GMT]
[Edited at 2025-05-12 23:37 GMT]
[Edited at 2025-05-12 23:38 GMT] ▲ Collapse
| | | | Lieven Malaise
بیلجیئم
Local time: 17:27
رکن (2020)
ڈچسےفرانسیسی
+ ...
Daryo wrote:
Want to trust confidential data to the kind of people who respond to being caught red-handed pirating other people's data by lobbying to make it legal when THEY do it? Count me out.
You do you. I'll do what I think is in my best interest. And as a business that is relying on legal agreements, among other things. I'll leave the panicking to people like you.
| | | | Daryo
برطانیہ
Local time: 16:27
انگریزیسےسربیائی
+ ...
| Adding and removing ... | May 22 |
Lieven Malaise wrote:
Daryo wrote:
Want to trust confidential data to the kind of people who respond to being caught red-handed pirating other people's data by lobbying to make it legal when THEY do it? Count me out.
You do you. I'll do what I think is in my best interest. And as a business that is relying on legal agreements, among other things. I'll leave the panicking to people like you.
Interesting - I wasn't aware than merely being aware of s.t. is a synonym for 'panicking' - plenty of dictionaries seem to have missed this! That should be added, and this probably removed from all and any Dictionary of Quotations for being pure panicking nonsense:
'Upton Sinclair said, it's difficult to get a man to understand something when his salary depends on his not understanding it.
| | |
|
|
|
| Hackers use AI too | May 22 |
bkking wrote:
DeepL (I'm talking about the Pro versions), is subject to the GDPR (General Data Protection Regulation) and the German Federal Data Protection Act (BDSG), which are extremely strict regarding Data Protection. They also highlight advanced encryption and compliance with regulations like ISO 27001 (information security management) and SOC 2 Type II (security, availability, and confidentiality).
[Edited at 2025-05-12 23:38 GMT]
Gjorgji Apostolovski wrote:
Facebook also stated the same, but it was breached.
And it's not just the software itself or some website, you know, there's a lot of ways to breach into someone's data.
So, you can't just say: I paid Pro license and now I can sleep calm.
Exactly! No company can promise that they won't be hacked. Remember, hackers use AI too, so it could even be that they would know/use the same GenAI code used by companies without companies knowing about it.
bkking wrote:
Even then, I find it to be good practice to never include personally identifiable information.
[Edited at 2025-05-12 23:38 GMT]
That's the best practice indeed. Do not trust the companies' speeches, because even paid versions could be hacked.
| | | |
I actually think it's very unlikely sensitive data would get hacked with DeepL Pro. They're bound by tough EU privacy laws (GDPR) and have major security certifications like ISO 27001, which means their security is constantly checked and improved. But the main thing? The translated text isn't stored after it's processed (in pro versions), so there's literally nothing for hackers to steal if a breach happened. That's the real key. Still, it's always smart to avoid typing in personal details, like... See more I actually think it's very unlikely sensitive data would get hacked with DeepL Pro. They're bound by tough EU privacy laws (GDPR) and have major security certifications like ISO 27001, which means their security is constantly checked and improved. But the main thing? The translated text isn't stored after it's processed (in pro versions), so there's literally nothing for hackers to steal if a breach happened. That's the real key. Still, it's always smart to avoid typing in personal details, like you say, that's good practice. Have a good one! ▲ Collapse
| | | | | Who will watch the watchmen? | May 26 |
Lieven Malaise wrote:
It's not about belief. It's business. They do a legally binding promise to refrain from doing certain things.
Blanket surveillance against law-abiding American citizens clearly violates the U.S. Constitution. Yet PRISM happened anyway. If governments readily broke the laws, why should I expect for-profit companies to stay true to their word?
And I still can't believe translators are confusing the definition of free as in libre vs. gratis. English is hard, I know.
[Edited at 2025-05-26 04:56 GMT]
| | | | | موضوع میں صفحات: < [1 2 3] > | To report site rules violations or get help, contact a site moderator: You can also contact site staff by submitting a support request » Poll: Does using AI in translation constitute breach of confidentiality? | Wordfast Pro |
|---|
Translation Memory Software for Any Platform
Exclusive discount for ProZ.com users!
Save over 13% when purchasing Wordfast Pro through ProZ.com. Wordfast is the world's #1 provider of platform-independent Translation Memory software. Consistently ranked the most user-friendly and highest value
Buy now! » |
| | Trados Business Manager Lite |
|---|
Create customer quotes and invoices from within Trados Studio
Trados Business Manager Lite helps to simplify and speed up some of the daily tasks, such as invoicing and reporting, associated with running your freelance translation business.
More info » |
|
| | | | X Sign in to your ProZ.com account... | | | | | |
Login to reply/comment 
